Biochemical and Biophysical Research Communications 493 (2017) 1356e1363

Contents lists available at ScienceDirect

Biochemical and Biophysical Research Communications

journal homepage: www.elsevier.com/locate/ybbrc

Age-related changes in the metabolic profiles of rat hippocampus,

medial prefrontal cortex and striatum
Lina Wati Durani a, Hamizah Shahirah Hamezah a, Nor Faeizah Ibrahim b,
Daijiro Yanagisawa a, Suzana Makpol b, Hanafi Ahmad Damanhuri b, Ikuo Tooyama a, *
a
Molecular Neuroscience Research Center, Shiga University of Medical Science, Seta Tsukinowa-cho, Otsu 520-2192, Japan
b
Department of Biochemistry, Faculty of Medicine, UKMMC, Universiti Kebangsaan Malaysia, Jalan Yaacob Latif, 56000 Cheras, Kuala Lumpur, Malaysia

a r t i c l e i n f o a b s t r a c t

Article history: We have recently shown that age-dependent regional brain atrophy and lateral ventricle expansion may
Received 19 September 2017 be linked with impaired cognitive and locomotor functions. However, metabolic profile transformation in
Accepted 29 September 2017 different brain regions during aging is unknown. This study examined metabolic changes in the hip-
Available online 29 September 2017
pocampus, medial prefrontal cortex (mPFC) and striatum of middle- and late-aged Sprague-Dawley rats
using ultrahigh performance liquid chromatography coupled with high-resolution accurate mass-
Keywords:
orbitrap tandem mass spectrometry. Thirty-eight potential metabolites were altered in hippocampus,
Aging
29 in mPFC, and 14 in striatum. These alterations indicated that regional metabolic mechanisms in lated-
Metabolomics
Hippocampus
aged rats are related to multiple pathways including glutathione, sphingolipid, tyrosine, and purine
Medial prefrontal cortex metabolism. Thus, our findings might be useful for understanding the complexity of metabolic mecha-
Striatum nisms in aging and provide insight for aging and health span.
© 2017 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction metabolomic perturbations in the hippocampus, medial prefrontal

cortex (mPFC), and striatum of middle-aged (MA) and late-aged
Brain aging is a progressive and complex multifactorial process (LA) rats using ultrahigh performance liquid chromatography
manifested by physiological and cognitive deterioration ultimately (UHPLC) coupled with high-resolution accurate mass (HRAM)-
leading to death [1]. It is a major risk factor for neurodegenerative orbitrap tandem mass spectrometry (MS/MS). This platform offers
diseases. Recently, brain metabolic profile abnormalities a convenient method for studying metabolic changes associated
throughout aging, including abnormal lipid metabolism, such as with brain aging with high precision, accuracy, and sensitivity [8].
dyshomeostasis of sphingolipid and myelin synthesis [2,3], have
been reported. Deficits in mitochondrial energy metabolism, anti- 2. Materials and methods
oxidant defense systems, neurotransmitter metabolism and amino
acid homeostasis have also been observed [3e5]. 2.1. Animals
The extent of change varies with brain regions. Thus, age-related
aberrations are highly confined to specific regions, including the All experimental procedures were approved by the Animal Care
hippocampus, frontal cortex and striatum [1,6]. In our previous and Use Committee of the Shiga University of Medical Science.
study of a rat aging model, we detected regional brain volumetric Forty-four male Sprague-Dawley rats (Clea Japan, Tokyo) were
changes and behavioral impairment in the late-aged rats [7]. housed in cages and maintained at room temperature under a 12-
However, little information is available about mechanisms under- hour light/dark cycle with food and water available ad libitum.
lying age-related cognitive impairments. Thus, it is important to The rats were divided into four age groups: 14 (n ¼ 14), 18 (n ¼ 13),
analyze not only regional brain atrophy but also metabolic changes 23 (n ¼ 7), and 27 (n ¼ 10) months. Behavioral evaluations and MRI
in different brain regions. Therefore, we examined regional measurements were performed as in our previous study [7]. After
the measurements, all rats were anesthetized with isoflurane to
remove their brains. Metabolites in different brain regions were
* Corresponding author. analyzed in four brains in each group. As shown in the previous
E-mail address: [email protected] (I. Tooyama). study, behavioral tests showed a big difference between middle-

https://doi.org/10.1016/j.bbrc.2017.09.164
0006-291X/© 2017 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
 L.W. Durani et al. / Biochemical and Biophysical Research Communications 493 (2017) 1356e1363
Fig. 1. Score plots of the OPLS-DA model showing a clear separation between groups. (AeC) ESI (þ) analysis. (DeF) ESI () analysis. Each point represents a single injection.

1357
1358 L.W. Durani et al. / Biochemical and Biophysical Research Communications 493 (2017) 1356e1363

aged (14 and 18 months) and late-aged (23 and 27 months) rats. unknown compounds (MW  RT) and the “Predict Compositions”
Therefore, they were divided into two groups (middle- and late- node to determine the possible elemental compositions. The pro-
aged) and metabolites were analyzed. Brain regions were cessing workflow also consisted of a library search against the
dissected using a brain matrix (Ted Pella, Inc., CA, USA), sectioned at mzCloud™ high-resolution accurate mass (HRAM) fragmentation
2.0e3.0 mm coronal slice intervals. The samples were transferred library, a mass list search against the built-in HRAM EFS library, and
to individual tubes, weighed, and stored at 80  C until assay. mapping to KEGG pathways. Mass tolerance on every node was set
at 5 ppm. The pre-processed data were exported to .xlsx files for
2.2. Sample preparation multivariate analysis.

The tissue samples were submerged in water-methanol- 2.5. Multivariate statistics

chloroform (2:1.5:1, v/v/v) and homogenized using a pestle and
glass beads. The mixture was sonicated and centrifuged at Peak areas data of the detected metabolites were subjected to
16,000  g for 10 min at 4  C. Subsequently, the tubes were multivariate statistical analysis using SIMCA-P™ software (version
transferred to a 20  C freezer and incubated overnight to allow 14.1, Umetrics AB, Umea, Sweden). Data were submitted to Pareto
residual chloroform to separate from the aqueous methanol phase. scaling to decrease the relative importance of larger values and for
The two liquid phases were transferred to separate 1.5 mL micro- logarithmic transformation to approximate a normal distribution.
centrifuge tubes and lyophilized at 50  C (Thermo Scientific Savant Unsupervised principal component analysis (PCA) was applied for a
SPD 1010 SpeedVac, USA). Only the methanol phase was used for preliminary evaluation of the quality of the data, followed by su-
the current analysis. The dry residue was reconstituted in cold pervised orthogonal partial least square-discriminant analysis
methanol (1:1, tissue weight/volume), diluted to a final volume of (OPLS-DA) for identifying the varied metabolites responsible for the
200 mL with the initial LC mobile phase, and filtered using a 2.0 mm separation between MA and LA groups. The model's quality was
membrane before analyzing. assessed by the R2 and Q2 values supplied by the software, which
provide information about goodness of fit and model predictive
2.3. Metabolomic analysis by UHPLC-MS/MS power, respectively. Potential metabolites were selected according
to a variable important in the projection (VIP) score above 1.0,
All solvents were LC-MS grade and were purchased from Fisher indicative of significant differences among groups. In addition,
Scientific (New Jersey, USA). Chromatography was performed on an MetaboAnalyst 3.0 software was used to conduct t-tests and fold
UltiMate 3000 Rapid Separation LC system coupled to a Thermo change analysis. False discovery rates (FDR) below 0.05 were
Scientific Q Exactive HF Orbitrap MS (Bremen, Germany). Chro- regarded as statistically significant. An S-plot of the OPLS-DA was
meleon Xpress system and Thermo Xcalibur™ MS data system calculated to visualize the variable influence in the model. Variables
were system controllers. Chromatographic separations were per- with a VIP value more than 1 and an FDR value less than 0.05 were
formed using a reversed-phase syncronis C18, 100  2.1 mm, 1.7 mm considered significant.
column (Thermo Scientific, MA, USA) at 55  C, with an injection
volume of 2 ml (n ¼ 3 injections for each sample). The auto-sampler
2.6. Metabolite identification
temperature was maintained at 0e10  C and the samples were
randomly injected. Solvents were delivered at a flow rate 0.45 ml/
Discriminant metabolites (VIP > 1.0 and significant by
min, using water (solvent A) and acetonitrile (solvent B), both
FDR < 0.05) detected by LC-MS profiling were identified using
containing 0.1% formic acid. The gradient elution program was set
experimental accurate mass by searching the Metlin library and
as follows: 0 min, 99.5% B; 5.5 min, 50% B; 6e13 min, 2% B;
Human Metabolome Database (HMDB) [9e11] with a mass accu-
13.1e15 min, 99.5% B. Positive and negative ionization modes were
racy of 5 ppm. These online databases are linked to KEGG (http://
employed under the following conditions: sheath gas flow rate of
www.genome.jp/kegg/), PubChem (https://pubchem.ncbi.nlm.nih.
20 arbitrary units (AU), auxiliary gas flow rate of 6 AU, sweep gas
gov/), LIPIDMAPS (http://www.lipidmaps.org/), and ChEBI
flow rate of 0 AU, capillary temperature 320  C, and spray voltage
(https://www.ebi.ac.uk/chebi/), and were used for further investi-
3.5 kV. Mass spectra were acquired using the full MS/dd-MS2 mode,
gation. Putative identifications were confirmed by comparing MS/
in which MS2 is triggered if an ion with high intensity is found in
MS spectra against the online mzCloud mass spectral database
full MS. The orbitrap mass spectrometer was operated in Fourier
linked to CD 2.0 software.
transform mass spectrometry (FTMS) full MS scan mode with a
high mass resolution of 60,000 and scan range of 1001000 m/z.
2.7. Metabolic pathway analysis
Moreover, the samples analyzed for mass fragmentation (FTMS,
high energy collision-trap dissociation (HCD), (20, 40, and 60
To determine and visualize perturbed metabolic pathways in
normalized collision energy (NCE)), MS2) with 1.5 m/z isolation
MA and LA groups, the identified metabolites were analyzed using
window and 15,000 FTMS. The MS was calibrated using Pierce LTQ
Metabolomic Pathway Analysis (MetPA) software. We selected the
ESI Positive Ion and Negative Ion Calibration Solution (Thermo
Rattus norvegicus library and used the “Fisher's Exact Test” and
Scientific). Quality control samples were prepared by mixing an
equal volume of each sample to assess the reproducibility and
reliability of the LC-MS/MS system. This pooled QC sample was Table 1
injected five times prior to individual sample analysis and after Statistical parameters of OPLS-DA models. A: number of latent components; R2:
every 12 sample runs. variance explained; Q2: variance predicted.

Hippocampus mPFC Striatum

2.4. Data pre-processing
A 1þ6 1þ5 1þ5
ESI (þ) R2 0.998 0.999 0.996
RAW data files were processed using Thermo Scientific Com- Q2 0.854 0.888 0.896
pound Discoverer™ 2.0 (CD 2.0) software with an untargeted A 1þ7 1þ6 1þ6
workflow (Fig. S1). This processing workflow uses the “Detect ESI () R2 0.999 0.999 0.999
Q2 0.917 0.884 0.935
Unknown Compounds” node to find chromatographic peaks of
 L.W. Durani et al. / Biochemical and Biophysical Research Communications 493 (2017) 1356e1363
Fig. 2. S-plot showing variables that contribute significantly to the differentiation between the MA (green) and LA groups (red). (AeC) ESI (þ) analysis. (DeF) ESI () analysis.

1359
1360 L.W. Durani et al. / Biochemical and Biophysical Research Communications 493 (2017) 1356e1363

Table 2
Potential metabolites for discrimination between MA and LA groups. *identified by mzCloud;**identified by Metlin; “C” identified by KEGG; “HMDB” identified by Human
Metabolome Database; “LM” identified by LipidMaps.

Metabolites ESI Identifier m/z Fold change

Hippocampus

4-Guanidinobutanoate þ C01035 146.09222 [

Butyrylcarnitine þ HMDB02013/LMFA07070054 232.15401 [
20 -Deoxymugineic acid þ C15485 305.13397 [
Bis-gamma-glutamylcystine þ C03646 499.11639 [
Oxoglutaric acid (a-ketoglutarate) þ HMDB00208 147.02852 [
L-gamma-glutamyl-L-isoleucine þ HMDB11170 261.14423 [
C17 Sphinganine þ LMSP01040003 288.28946 [
PI-Cer(d18:0/22:0) þ LMSP03030009 866.64746 [
PC(O-20:1(11Z)/20:4(8Z,11Z,14Z,17Z)) þ LMGP01020274 822.64044 [
N-Ribosylhistidine þ HMDB02089 288.11838 [
Deoxy-5-methylcytidylate þ C03495 360.03415 [
(a-D-mannosyl)2-b-D-mannosyl-N-acetylglucosamine þ HMDB06537 708.25562 [
Tauropine þ C01616 198.0428 [
Sphinganine þ C00836 302.30505 [
S-Methylglutathione þ C11347 322.10596 [
UMP þ C00105 325.04242 [
Phosphoribosyl-AMP þ C02741 560.07916 [
Cer(d18:0/16:0) þ HMDB11760 540.53412 [
D-Fructose 1,6-bisphosphate þ C00354 341.00284 [
PG(16:0/0:0) þ LMGP04050008 507.27133 [
Galabiosylceramide (d18:1/18:0) þ HMDB04834 890.65680 Y
N,N-dimethyl-Safingol þ LMSP01080056 330.33563 [
20 -Deoxycytidine 50 -monophosphate (dCMP) e 136* 306.04968 [
L-Glutathione (reduced) e 472* 306.07635 [
L-Histidinol e 475* 140.0819 [
D-Sedoheptulose 7-phosphate e 1503* 289.0329 [
4-Hydroxy-N-(2-diethylaminoethyl)benzamide e HMDB35174 418.09366 [
2,3-Dihydroxybenzoylserine e C04204 240.05127 [
ADP-L-glycero-beta-D-manno-heptose e C06398 618.08539 [
Tauropine (N2-(D-1-Carboxyethyl)taurine) e C01616 196.02794 [
beta-Citryl-L-glutamate e C20775 320.0624 [
PC(23:1/23:1) e LMGP01011141 924.74298 [
N-Acetylneuraminate 9-phosphate e C06241 388.06485 [
Sedoheptulose 1,7-bisphosphate e C00447 368.99902 [
gamma-L-Glutamyl-L-cysteine e C04544 320.09201 [
30 -sulfogalactosylceramide e HMDB00024 906.63623 Y
L-Fucose 1-phosphate e C02985 243.0273 [
6-Phospho-D-gluconate e C00345 275.0173 [
mPFC
Triethanolamine þ 750* 150.11205 [
L-Glutathione (reduced) þ 472* 308.09006 Y
6-Acetamido-2-oxohexanoate þ C05548 188.09131 [
5-Acetamidopentanoate þ C03087 182.07864 [
(R)-S-Lactoylglutathione þ C03451 380.11124 Y
S-Adenosyl-L-homocysteine þ C00021 407.10986 [
Cysteinylglycine þ 363* 179.048 [
N-a-L-Acetyl-arginine þ 515* 217.12921 [
4-Hydroxyphenylacetaldehyde þ C03765 137.05942 [
1-deoxy-sphinga-6Z,9Z,12Z,15Z-tetraenine þ LMSP01080036 278.24716 [
10-nitro-9E-octadecenoic acid þ LMFA01120003 350.22928 [
L-2-Aminoadipate þ C00956 162.07568 Y
gamma-L-Glutamyl-L-cysteine þ C00669 251.06903 Y
Methionyl-Tyrosine þ HMDB28985 313.12158 [
Indole þ C00463 118.06512 [
PC(8:0/0:0) þ LMGP01050066 384.21384 [
Dopamine þ 4* 154.086 Y
GDP-3,6-dideoxy-D-galactose þ C05136 574.09351 Y
Phe-Tyr-OH e 65175** 435.11844 Y
L-Hexanoylcarnitine e HMDB00756 258.17072 [
O-Acetylcarnitine e C02571 202.10759 [
4-(4-Deoxy-alpha-D-gluc-4-enuronosyl)-D-galacturonate e C06118 351.056 Y
S-Lactoylglutathione e 597* 378.09647 Y
GDP-3,6-dideoxy-D-galactose e C05136 572.07941 Y
L-Glutathione (reduced) e 472* 306.07559 Y
Dephospho-CoA e C00882 686.14142 Y
3-Methylglutaric acid e 211* 145.04945 [
Prostaglandin D2-d9 e 96462** 360.27515 [
Suberic acid e 1393* 173.08084 [
Striatum
Hypoxanthine þ 441* 137.04547 [
Adenylsuccinic acid þ HMDB00536 464.08044 [
Inosine-50 -monophosphate (IMP) þ 452* 349.05392 [
 L.W. Durani et al. / Biochemical and Biophysical Research Communications 493 (2017) 1356e1363 1361

Table 2 (continued )

Metabolites ESI Identifier m/z Fold change

Hippocampus

L-Citrulline þ HMDB00904 176.10257 [

Adenylosuccinic acid e C03794 462.06598 [
2,5-Dihydroxybenzaldehyde e 131* 137.02319 Y
N6-Acetyl-LL-2,6-diaminoheptanedioate e C04390 231.09802 Y
1-(50 -Phosphoribosyl)-5-formamido-4-imidazolecarboxamide e C04734 365.04996 Y
IMP e C00130 347.03891 [
Adenine e 296* 134.0459 Y
Xanthine e 781* 151.0249 Y
Adenosine e 297* 266.08911 Y
1-(sn-Glycero-3-phospho)-1D-myo-inositol @ glycerophosphoinositol e HMDB11649 333.05853 [
D-Glucosamine 6-phosphate e C00352 258.03775 [

“Relative-Betweenness Centrality” algorithms for pathway enrich- multivariate statistical analysis could identify and differentiate
ment and pathway topological analyses, respectively. Pathways brain metabolites between MA and LA rats. Interestingly, not all
with impact-values above 0.1 are considered most relevant [12]. metabolites displayed significant changes consistently throughout
the three brain regions. The results suggest that the metabolic al-
3. Results terations in brain aging may be related to brain functions in each
region. For example, the hippocampus and prefrontal cortex are
In total, 1274 accurate masses in the hippocampus using the responsible for cognitive function, including learning and memory
positive electrospray ionization mode (ESI (þ)) and 990 were [16,17], while the striatum plays an important role in movement
detected using the negative electrospray ionization mode (ESI ()) control [18].
mode. Further, 1443 and 1751 accurate masses were detected by ESI Pathways analysis revealed that the most affected pathway in
(þ) and ESI () in the mPFC, respectively, and 1330 and 979, the hippocampus and mPFC was glutathione (GSH) metabolism.
respectively, in the striatum. The preprocessed data were subjected GSH, a key low molecular weight antioxidant, is considered the
to multivariate analysis for sample classification. The Hotelling's T2- main indicator of the overall cellular redox state [19]. The redox
range plot showed no significant outliers. As already described state of tissues shifts toward a more oxidized state during aging.
[13e15], preliminary data evaluation using PCA could not Previous studies have reported reduced GSH function in the aged
discriminate between groups (Fig. S2). Therefore, we applied OPLS- animal brain [20e22], leading to an increased susceptibility to
DA to the same data set to sharpen the separation and to identify oxidative injury. These studies also proposed a regional variation in
metabolites contributing to the differences between groups. GSH concertation. However, most of them used enzyme assay and
OPLS-DA score plots displayed a distinct separation between MA only a few studies applied metabolomic platforms. We observed
and LA groups (Fig. 1), indicating their different metabolic profiles. that GSH, L-cysteinyl glycine and L-g-glutamylcysteine were
High R2 and Q2 values demonstrated OPLS-DA model robustness. reduced in the mPFC, indicating disturbed GSH metabolism.
Table 1 shows model quality. Cumulative values of R2 and Q2 were Decreased GSH synthesis rate or increased utilization to combat
all above 0.8, indicating excellent models. Fifty-nine significant oxidative stress may cause this. Conversely, reduced glutathione
metabolites were detected in the hippocampus using the ESI (þ) (GSSH), L-g-glutamylcysteine, bis-g-glutamylcysteine, and S-
mode and 53 using the ESI () mode, while 166 and 87 significant methylglutathione were increased in the hippocampus of LA rats,
variables were detected by ESI (þ) and ESI () in mPFC, respec- suggesting improved GSH metabolism. Previous metabolomic
tively, and 12 and 50, respectively, in the striatum. S-plots of the studies have reported significantly increased GSH in the brain of
OPLS-DA were used to highlight the significant metabolites (Fig. 2). Parkinson's disease and Alzheimer's disease patients [23,24], while
The selected metabolites have a high covariance combined with a a slight increase was also noticed in 26-month-old compared to 12-
high correlation, producing more reliable results. Ultimately, the month-old mice [20]. Overall, increased GSH in the hippocampus
analysis showed 36 upregulated metabolites and two down- may play a protective role.
regulated metabolites in the hippocampus, while 14 and 15 me- Taurine and hypotaurine metabolism was the second most
tabolites in mPFC and 8 and 6 in the striatum were upregulated and affected pathway in the hippocampus. A higher tauropine (N2-(D-
downregulated, respectively. 1-Carboxyethyl)taurine) level was observed. Tauropine is the
Table 2 lists the potential metabolites identified by UHPLC- product of tauropine dehydrogenase, which catalyzes a chemical
HRAM-orbitrap MS/MS. A metabolomics pathway analysis con- reaction involving taurine and pyruvate as substrates. Taurine is a
structed 14 metabolic pathways in the hippocampus, 9 in the mPFC, sulfur-containing, antioxidant amino acid. It is abundant in the
and 5 in the striatum (Fig. 3). Among these, the high impact-value developing brain and is involved in neurogenesis [25]. Previous
pathways were glutathione metabolism (impact-value: 0.439), metabolomics studies have reported taurine involvement in brain
taurine and hypotaurine metabolism (impact-value: 0.429), and aging [3,26,27] and neurodegenerative diseases [28,29], while
sphingolipid metabolism (impact-value: 0.281) in the hippocam- chronic taurine supplementation preempts age-related cognitive
pus; glutathione metabolism (impact-value: 0.447), tyrosine function decline, mainly through GABAergic system alterations
metabolism (impact-value: 0.151), and sphingolipid metabolism [30].
(impact-value: 0.143) in the mPFC; and purine metabolism Metabolites associated with sphingolipid metabolism were
(impact-value: 0.238) in the striatum. identified in the hippocampus and mPFC: 2-3-dihydroxyl-ben-
zoylserine (a serine derivative), C17-sphinganine, dihy-
drosphingosine (C18-sphinganine), PI-ceramide (d18:0/22:0), and
4. Discussion N,N-dimethyl-safingol were increased; 3-O-sulfogalactosylcer-
amide (d18:1/24:0) was decreased in the hippocampus; and 1-
In this study, UHPLC-HRAM-orbitrap MS/MS coupled with
1362 L.W. Durani et al. / Biochemical and Biophysical Research Communications 493 (2017) 1356e1363

deoxy-sphinga-6Z,9Z,12Z,15Z-tetraenine (a sphingoid base) was

increased in the mPFC. Sphingolipids are major cell membrane
components mostly found in the brain, and participate in diverse
cellular functions. Sphingolipids are synthesized de novo in the
endoplasmic reticulum, generating ceramide, which is then trans-
ported to the golgi complex to produce complex sphingolipids like
sphingomyelin and glycosphingolipids. Sphinganines accumulation
in the hippocampus of LA rats suggests defected ceramide synthase
that converts sphinganine to dihydroceramide. Yoo et al. [31] re-
ported high free sphinganine concentration due to perturbed

Fig. 3. Pathway analysis overview showing altered metabolic pathways. The color and size of each circle is based on p-value and pathway impact-value, respectively.
sphinganine metabolism. Sulfatide (3-O-sulfogalactosylceramide
(d18:1/24:0)) is a sulfate-containing glycosphingolipid. Low sulfa-
tide levels are observed in the hippocampus of LA rats, probably
caused by a failure in ceramide transportation to the golgi complex.
Sulfatide is abundant in the myelin sheath, is essential for myelin
function, and regulates oligodendrocyte differentiation [32]. Our
finding is supported by previous studies showing that brain sulfa-
tide dysregulation and myelin sulfatide content loss are observed
with aging. These alterations may be significant risk factors for
behavioral deficits observed in normal aging [33e35] and age-
related neurological disorders [36,37].
Purine and pyrimidine nucleotides are primary energy carriers,
subunits of DNA and RNA, and precursors for nucleotide cofactor
synthesis, such as nicotinamide adenine dinucleotide and S-ade-
nosylmethionine. Pathway analysis revealed that purine meta-
bolism was the most altered pathway in the striatum. 1-(50 -
Phosphoribosyl)-5-formamido-4-imidazolecarboxamide (FAICAR),
adenosine, adenine, and xanthine were downregulated, while
inosine monophosphate (IMP), adenylosuccinate and hypoxan-
thine were upregulated in the brains of LA rats compared with MA
rats. Recently, brain metabolome analysis in male rats across the
lifespan demonstrated downregulated purine metabolites in aged
animals [3]. Furthermore, purine metabolism deregulation has
been reported in an animal model and in Alzheimer's disease pa-
tients [38,39]. Adenosine has gained attention among researchers
for its function as a neuromodulator and neuroprotector. It regu-
lates synaptic transmission and neuronal excitability in the central
nervous system via specific adenosine receptors [40,41]. Xanthine
and hypoxanthine are oxidation products of purine metabolism
and have been detected in the brain tissue of neurodegenerative
disease patients and animal models [42]. Previous studies sug-
gested that purine metabolism was implicated in the mechanisms
underlying neurodegenerative disorders.
Declined brain dopamine activity with age and age-related
diseases is well-documented, and is associated with age-related
locomotor impairment [43,44]. Dopamine loss corresponds to
dopaminergic neurons degeneration in the substantia nigra and
striatum [45e47]. Interestingly, dopamine and 4-
hydroxylphenylacetaldehyde (a tyrosine metabolism byproduct)
were reduced in the mPFC. The mPFC receives dopaminergic inputs
from the substantia nigra as a meso-cortical dopaminergic
pathway. The pathway is implicated in some emotional changes
and cognitive deficits. mPFC region dysregulation with dopamine
reduction was shown by earlier studies [48,49].
In conclusion, we indicate multiple metabolic pathways per-
turbed during aging. The metabolic changes were different be-
tween brain regions. High impact-value pathways were glutathione
metabolism, taurine and hypotaurine metabolism, and sphingoli-
pid metabolism in the hippocampus; glutathione metabolism,
tyrosine metabolism, and sphingolipid metabolism in the mPFC;
and purine metabolism in the striatum. The metabolic changes may
reflect functional changes, including cognitive impairments with
aging.
 L.W. Durani et al. / Biochemical and Biophysical Research Communications 493 (2017) 1356e1363 1363

Acknowledgements [22] I. Rebrin, M.J. Forster, R.S. Sohal, Effects of age and caloric intake on gluta-
thione redox state in different brain regions of C57BL/6 and DBA/2 mice, Brain
Res. 1127 (2007) 10e18.
This work was supported by Grant-in-Aids from the Japan So- [23] M. Bogdanov, W.R. Matson, L. Wang, et al., Metabolomic profiling to develop
ciety for Promotion of Science (KAKENHI 17H03560 and 17K01355) blood biomarkers for Parkinson's disease, Brain 131 (2008) 389e396.
and the Ministry of Education Malaysia (LRGS/BU/2012/UKM-UKM/ [24] R. Kaddurah-Daouk, H. Zhu, S. Sharma, et al., Alterations in metabolic path-
ways and networks in Alzheimer's disease, Transl. Psychiatry 3 (2013) e244.
K/04). [25] E. Gebara, F. Udry, S. Sultan, et al., Taurine increases hippocampal neuro-
genesis in aging mice, Stem Cell Res. 14 (2015) 369e379.
Transparency document [26] A.P. Sobolev, L. Mannina, M. Costanzo, et al., Age-related changes in skeletal
muscle composition: a pilot nuclear magnetic resonance spectroscopy study
in mice, Exp. Gerontol. 92 (2017) 23e27.
Transparency document related to this article can be found [27] Q.-L. Wan, X. Shi, J. Liu, et al., Metabolomic Signature Associated with
online at https://doi.org/10.1016/j.bbrc.2017.09.164. Reproduction-regulated Aging in Caenorhabditis elegans, vol. 9, Aging, Albany
NY, 2017, p. 447.
[28] S.F. Graham, C. Holscher, B.D. Green, Metabolic signatures of human Alz-
Appendix A. Supplementary data heimer's disease (AD):^ sup 1^ H NMR analysis of the polar metabolome of
post-mortem brain tissue, Metabolomics 10 (2014) 744.
[29] L.M. Villeneuve, P.R. Purnell, M.D. Boska, et al., Early expression of Parkinson's
Supplementary data related to this article can be found at disease-related mitochondrial abnormalities in PINK1 knockout rats, Mol.
https://doi.org/10.1016/j.bbrc.2017.09.164. Neurobiol. 53 (2016) 171.
[30] A. El Idrissi, Taurine improves learning and retention in aged mice, Neurosci.
Lett. 436 (2008) 19e22.
References
[31] H.-S. Yoo, W.P. Norred, J. Showker, et al., Elevated sphingoid bases and
complex sphingolipid depletion as contributing factors in fumonisin-induced
[1] R. Peters, Ageing and the brain, Postgrad. Med. J. 82 (2006) 84e88. cytotoxicity, Toxicol. Appl. Pharmacol. 138 (1996) 211e218.
[2] V. Paban, F. Fauvelle, B. Alescio-Lautier, Age-related changes in metabolic [32] J.M. Boggs, Role of Galactosylceramide and Sulfatide in Oligodendrocytes and
profiles of rat hippocampus and cortices, Eur. J. Neurosci. 31 (2010) CNS Myelin: Formation of a Glycosynapse, Glycobiology of the Nervous Sys-
1063e1073. tem, Springer, 2014, pp. 263e291.
[3] X. Zheng, T. Chen, A. Zhao, et al., The brain metabolome of male rats across the [33] G. Bartzokis, J.L. Cummings, D. Sultzer, et al., White matter structural integrity
lifespan, Sci. Rep. 6 (2016). in healthy aging adults and patients with Alzheimer disease: a magnetic
[4] J. Ivanisevic, K.L. Stauch, M. Petrascheck, et al., Metabolic Drift in the Aging resonance imaging study, Arch. Neurol. 60 (2003) 393e398.
Brain, vol. 8, Aging, Albany NY, 2016, p. 1000. [34] N.A. Crivello, S.L. Casseus JW, et al., Age- and brain region-specific effects of
[5] M. Jove
, M. Portero-Otín, A. Naudí, et al., Metabolomics of human brain aging dietary vitamin K on myelin sulfatides, J. Nutr. Biochem. 21 (2010)
and age-related neurodegenerative diseases, J. Neuropathol. Exp. 73 (2014) 1083e1088.
640e657. [35] L. Svennerholm, K. Bostro € m, B. Jungbjer, et al., Membrane lipids of adult hu-
[6] N. Raz, F. Gunning-Dixon, D. Head, et al., Aging, sexual dimorphism, and man brain: lipid composition of frontal and temporal lobe in subjects of age
hemispheric asymmetry of the cerebral cortex: replicability of regional dif- 20 to 100 years, J. Neurochem. 63 (1994) 1802e1811.
ferences in volume, Neurobiol. Aging 25 (2004) 377e396. [36] T. Takahashi, T. Suzuki, Role of sulfatide in normal and pathological cells and
[7] H.S. Hamezah, L.W. Durani, N.F. Ibrahim, et al., Volumetric changes in the tissues, J. Lipid Res. 53 (2012) 1437e1450.
aging rat brain and its impact on cognitive and locomotor functions, Exp. [37] X. Han, D.M. Holtzman, D.W. McKeel, et al., Substantial sulfatide deficiency
Gerontol 99 (2017) 69e79. and ceramide elevation in very early Alzheimer's disease: potential role in
[8] E.I. Hamelin, W. Bragg, R.L. Shaner, et al., Comparison of high resolution and disease pathogenesis, J. Neurochem. 82 (2002) 809e818.
tandem mass spectrometry for the analysis of nerve agent metabolites in [38] R. Gonz
alez-Domínguez, T. García-Barrera, J. Vitorica, et al., Metabolomic
urine, Rapid Commun, Mass Spectrom. 27 (2013) 1697e1704. screening of regional brain alterations in the APP/PS1 transgenic model of
[9] D.S. Wishart, T. Jewison, A.C. Guo, et al., HMDB 3.0ethe human metabolome Alzheimer's disease by direct infusion mass spectrometry, J. Pharm. Biomed.
database in 2013, Nucleic Acids Res. 41 (2013) D801eD807. Anal. 102 (2015), 425435.
[10] D.S. Wishart, C. Knox, A.C. Guo, et al., HMDB: a knowledgebase for the human [39] B. Ansoleaga, M. Jove
, A. Schlüter, et al., Deregulation of purine metabolism in
metabolome, Nucleic Acids Res. 37 (2009) D603eD610. Alzheimer's disease, Neurobiol. Aging 36 (2015) 68e80.
[11] D.S. Wishart, D. Tzur, C. Knox, et al., HMDB: the human metabolome database, [40] R. Cunha, Adenosine as a neuromodulator and as a homeostatic regulator in
Nucleic Acids Res. 35 (2007) D521eD526. the nervous system: different roles, different sources and different receptors,
[12] R. Gonzalez-Dominguez, T. Garcia-Barrera, J. Vitorica, et al., Metabolomic Neurochem. Int. 38 (2001) 107e125.
screening of regional brain alterations in the APP/PS1 transgenic model of [41] A. Rahman, The role of adenosine in Alzheimer's disease, Curr, Neuro-
Alzheimer's disease by direct infusion mass spectrometry, J. Pharm. Biomed. pharmacol 7 (2009) 207e216.
Anal. 102 (2015) 425e435. [42] C. Gonzalez-Riano, A. Garcia, C. Barbas, Metabolomics studies in brain tissue: a
[13] R. Gonzalez-Dominguez, T. Garcia-Barrera, J. Vitorica, et al., Deciphering review, J. Pharm. Biomed. Anal. 130 (2016) 141e168.
metabolic abnormalities associated with Alzheimer's disease in the APP/PS1 [43] N.D. Volkow, J. Logan, J.S. Fowler, et al., Association between age-related
mouse model using integrated metabolomic approaches, Biochimie 110 decline in brain dopamine activity and impairment in frontal and cingulate
(2015) 119e128. metabolism, Am. J. Psychiatry 157 (2000) 75e80.
[14] R.J. Mishur, S.L. Rea, Applications of mass spectrometry to metabolomics and [44] C. Moinard, L. Tliba, J. Diaz, et al., Citrulline stimulates locomotor activity in
metabonomics: detection of biomarkers of aging and of age-related diseases, aged rats: implication of the dopaminergic pathway, Nutrition 38 (2017)
Mass Spectrom. Rev. 31 (2012) 70e95. 9e12.
[15] M. Wong, J.K. Lodge, A metabolomic investigation of the effects of vitamin E [45] O. Hornykiewicz, Biochemical aspects of Parkinson's disease, Neurology 51
supplementation in humans, Nutr. Metab. (Lond) 9 (2012) 110. (1998) S2eS9.
[16] D.R. Euston, A.J. Gruber, B.L. McNaughton, The role of medial prefrontal cortex [46] C.U. Von Linstow, M. Severino, A. Metaxas, et al., Effect of aging and Alz-
in memory and decision making, Neuron 76 (2012) 1057e1070. heimer's disease-like pathology on brain monoamines in mice, Neurochem.
[17] M. Good, Spatial memory and hippocampal function: where are we now? Int. 108 (2017) 238e245.
Psicolo
gica 23 (2002) 109e138. [47] K. Kuter, Ł. Olech, U. Głowacka, Prolonged dysfunction of astrocytes and
[18] A.V. Kravitz, A.C. Kreitzer, Striatal mechanisms underlying movement, rein- activation of microglia accelerate degeneration of dopaminergic neurons in
forcement, and punishment, Physiology 27 (2012) 167e177. the rat substantia nigra and block compensation of early motor dysfunction
[19] I. Rebrin, S. Kamzalov, R.S. Sohal, Effects of age and caloric restriction on induced by 6-OHDA, Mol. Neurobiol. (2017) 1e18.
glutathione redox state in mice, Free Radic. Biol. Med. 35 (2003) 626e635. [48] N.D. Volkow, R.C. Gur, G.J. Wang, et al., Association between decline in brain
[20] T.S. Chen, J.P. Richie Jr., C.A. Lang, The effect of aging on glutathione and dopamine activity with age and cognitive and motor impairment in healthy
cysteine levels in different regions of the mouse brain, Proc. Soc. Exp. Biol. individuals, Am. J. Psychiatry 155 (1998) 344e349.
Med. 190 (1989) 399e402. [49] J.D. Cohen, T.S. Braver, J.W. Brown, Computational perspectives on dopamine
[21] X. Chen, C. Xie, L. Sun, et al., Longitudinal metabolomics profiling of Parkin- function in prefrontal cortex, Curr. Opin. Neurobiol. 12 (2002) 223e229.
son's disease-related a-synuclein A53T transgenic mice, PloS One 10 (2015)
e0136612.