International Journal of Mass Spectrometry 418 (2017) 1–5

Contents lists available at ScienceDirect

International Journal of Mass Spectrometry

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

Non-targeted sportomics analyses by mass spectrometry to

understand exercise-induced metabolic stress in soccer players
Eduardo Prado a,b , Gustavo H.M.F. Souza c , Marcelle Pegurier b,d , Camila Vieira b ,
Abelardo Barbosa Moreira Lima-Neto b,e , Marcio Assis f , Maria Izabel Florindo Guedes e ,
Maria Gabriela Bello Koblitz b,g , Mariana Simões Larraz Ferreira b,g ,
Andrea Furtado Macedo b,h , Altamiro Bottino i , Adriana Bassini b,c , L.C. Cameron b,c,∗
a
Laboratory for Research in Physical Exercise and Metabolism, Federal University of Alagoas – Av. Lourival Melo Mota, S/N, Tabuleiro do Martins
57072-970, Maceió, AL, Brazil
b
Laboratory of Protein Biochemistry – Federal University of State of Rio de Janeiro, Av. Pasteur, 296, Urca, 22290-250, Rio de Janeiro, Brazil
c
MS Applications & Development Laboratory, HRMS Health Sciences Department, Waters Corporation, São Paulo, Brazil
d
Department of Biochemistry and Sportomics, Olympic Laboratory, Brazil Olympic Committee, Brazil
e
Laboratory of Biotechnology and Molecular Biology, State University of Ceará, Brazil
f
Fluminense Football Club, Brazil
g
Nutritional Biochemistry Center, Federal University of State of Rio de Janeiro, Brazil
h
Integrated Laboratory of Plant Biology, Department of Botany, Institute of Biosciences, Federal University of State of Rio de Janeiro, Brazil
i
Sociedade Esportiva Palmeiras, Brazil

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

Article history: We have been using “-Omics” sciences with classic laboratory analyses to understand the systemic
Received 25 August 2016 metabolic and signaling changes induced by sport and exercise. We called this approach Sportomics. Our
Received in revised form 18 January 2017 samples were collected in situ either during competitions or training to mimic the genuine challenges
Accepted 1 February 2017
and conditions faced during sports. Non-targeted analysis (NTA) has opened the door to a new era of
Available online 13 February 2017
high-throughput exercise-induced metabolic research. In the present study, 30 male semi-professional
soccer players were observed for two subsequent days. Both blood and urine samples were collected
Keywords:
immediately pre-match and after matches. The most up-regulated prominent molecules were fatty acyls,
metabolomics
fatigue
carboxylic acids and derivatives, steroids and steroid derivatives. The most down-regulated molecules
soccer were fatty acyls, carboxylic acids and derivatives, as well as benzene and substituted derivatives. After
metabolite identification and determining which metabolites were up- or down-regulated, we took the
metabolites and grouped them into classes to examine the metabolic pathways involved with purine
metabolism and to investigate hyperammonemia. To follow-up on our findings in urine, we used point-
of-care instrument analysis to measure capillary blood metabolites. Glucose significantly increased by
35%, whereas urate increased by 16% and uremia by 17% without any changes in creatinemia. In the
present study, we showed that hypoxanthine and related metabolites were up-regulated in urine after
a soccer match, which suggested that AMP deamination was increased. In this study, we demonstrated
several results through urine non-target mass spectrometry (NTMS) to understand exercise-induced
changes during a soccer match using a Sportomics approach. These data together demonstrated that
during a soccer game, there was an increase in ATP use provided by ADP synthesis via myokinase. Our
data may show that the use of urinary metabolomics can be a less invasive way to follow the metabolism
of athletes during exercise. We demonstrated that the use of NTMS may be useful for future studies that
aim to design holistic interventions for improving athletic performance.
© 2017 Elsevier B.V. All rights reserved.

1. Introduction

∗ Corresponding author at: Laboratory of Protein Biochemistry – Federal Univer- Recently, we assembled a new concept in metabolic studies and
sity of State of Rio de Janeiro, Av. Pasteur, 296, Urca, Rio de Janeiro, 22290-250, exercise science called Sportomics through the use of “-Omics” sci-
Brazil. ences with classic laboratory analyses to understand the systemic
E-mail addresses: cameron@unirio.br, lccameron@me.com (L.C. Cameron).

http://dx.doi.org/10.1016/j.ijms.2017.02.002
1387-3806/© 2017 Elsevier B.V. All rights reserved.
2 E. Prado et al. / International Journal of Mass Spectrometry 418 (2017) 1–5
Fig. 1. Experiment design. Urine samples from young male semi-professional soccer players collected at two different time points, namely pre-match and post-match, were
tested.
metabolic and signaling changes induced by participation in sport.
Using this approach, the samples were collected in situ either dur-
ing competitions or training to mimic the genuine challenges and
conditions faced during participation in sports [3].
We believe that this approach can offer interdisciplinary con-
nections among fields to reduce the barriers between sport trainers
and scientists, including translational science, to improve athletic
performance significantly. We propose that in-field metabolic anal-
yses are better for understanding, supporting and training elite
athletes.
As with any “-Omics” science, Sportomics requires analyti-
cal techniques with great computational capability to manage
large amounts of data. The use of non-targeted analysis by
mass spectrometry (NTMS) is considered the “gold medal” cham-
pion approach to holistically understand an in vivo appraisal of
proteomic and metabolomic changes induced by exercise. NTA
opened the gate to a new era of high-throughput exercise-induced
metabolic research by delivering a better understanding of the inte-
gration of biological processes and extending our knowledge of
exercise to systems biology [5,10,1].
In this study, we obtained several results through urine NTMS to
understand exercise-induced changes during a soccer match using
a Sportomics approach. We also described changes such as purine
and pyrimidine metabolism, which can be utilized to recognize
exercise-induced hyperammonemia. Additionally, we conducted
this investigation to comprehend the metabolic changes that occur
in athletes undergoing exercise from a practical perspective.
2. Methods
2.1. Subjects
Thirty male semi-professional soccer players (18-20 years old)
from a team affiliated with the Confederação Brasileira de Futebol
(CBF, Brazilian Soccer Confederation) participated in this study as
volunteers. The players were healthy and did not have detectable
diseases. They were evaluated clinically twice per year. The subjects
were informed previously about the study, and written informed
consent was obtained from each one. All of the procedures were
performed according to the ethical standards of the Ethics Commit-
tee for Human Research at the Federal University of the State of Rio
de Janeiro (117/2007, renewed in 2011) and met the requirements
for regulating research on human subjects (Health National Coun-
cil, Brazil, 1996). The subjects were tested during two subsequent
days (n = 30).
2.2. Experimental Design, Sample Collection and Preparation
Five different urinary samples were collected the night before
the match (0 min), in the fasting state on the morning of the
match (720 min), immediately prior to the match (1080 min), at
the half-time break (1125 min) and after the match (1185 min).
We compared two time points, including the pre-match and post-
match times (Fig. 1).
After collection, urine was used as a biological matrix, and the
samples were immediately transferred to a dry ice cooler and were
transported to the Laboratory of Protein Biochemistry (LBP) located
at the Federal University of the State of Rio de Janeiro (UNIRIO).
Later, the samples were stored in an ultra-low temperature
freezer (–80 ◦ C) until they were prepared for ultra-pressure liq-
uid chromatography followed by alternating low- and high-energy
multiplexed MS/MS (UPLC-MSE ) injections. Mass spectrometry
analyses were further performed at a starting volume of 700 ␮L
of raw urine samples that were thawed to room temperature. Raw
urine samples were then centrifuged at 10,000 × g for 30 minutes at
4 ◦ C, and the supernatants were perfused through a dialysis mem-
brane with a 3,000 Da molecular weight cutoff (MWCO) (Amicon,
Merck Millipore, Germany). The filtrate was desalted using a solid-
®
phase hydrophilic-lipophilic-balanced extraction cartridge (Oasis
HLB, Waters Corporation, USA). The samples were concentrated
using a SpeedVac Plus (Model: SC110A, ThermoSavant, USA) and
were reconstituted in solvent solution containing 3% acetonitrile
and 0.1% formic acid in Milli-Q pure water. Finally, the samples
were transferred to a UPLC auto-sampler vial (Waters Corporation,
USA).
2.3. Point-of-care analysis
For comparison, blood was collected from seven athletes
immediately prior to the match (1080 min) and after the match
(1185 min) for measurement of glucose, urea, urate and creatinine.
The venous blood was collected and immediately centrifuged for
10 min at 3,000 × g to separate the plasma. The samples were stored
at −20 ◦ C. Each plasma sample of the players was thawed and pipet-
ted into General Chemistry 13 and MetLyte 8 discs to be analyzed
®
by Piccolo Xpress .
2.4. UPLC-MSE method
UPLC-MSE data were acquired in an ultra-high-performance liq-
uid chromatography system (Acquity UPLC I-Class, Waters, USA)
coupled to an ESI (+) Qq-oaTOF mass spectrometer (Xevo G2-S
Q-Tof, Waters, UK). A total of 10 ␮L of each sample was injected,
and the separation was performed on an ACQUITY UPLC CSH C18
Column, 130 Å, 1.7 ␮m, 2.1 mm × 50 mm conditioned at 40 ◦ C. The
mobile phases were 0.1% formic acid (pump A) and 0.1% formic
acid in acetonitrile (pump B), and the flow rate was 900 ␮L min−1 .
The gradient method was programmed to achieve maximum sep-
aration performance as follows: initial condition 3% B (pump B),
2.37 min 35% B, 4.37 min 85% B, 5.37 min 85% B, and 6.37 min 3% B
with a total run time of 8.37 min and a calculated percent B/column
volume (cV) factor of 2.6%B/cV. The sample tray temperature was
 E. Prado et al. / International Journal of Mass Spectrometry 418 (2017) 1–5
Fig. 2. Metabolic pathways of purine metabolism. The color green represents up-regulation, and the color red represents down-regulation.
defined at 8 ◦ C. The mass spectrometry method and conditions
were adjusted, including alternating the continuous ion current
with low- and high-energy multiplexed MS/MS mode (MSE ) that
could achieve a collision energy ramp that was set to 10–30 eV.
The MS instrument was controlled with the MassLynx V4.1 soft-
ware package (Waters Corporation, UK). The scanning mass range,
quadrupole profile and instrument calibration were set to transmit
the ion current from m/z 50 to 1000. The source conditions were
tuned as follows: capillary at 3 kV, sampling cone (skimmer) set
to 15 V, stepwave source offset of 30 V, cone gas flow (curtain gas)
set to 50 L h−1 , desolvation gas flow 802 L h−1 , source temperature
set to 100 ◦ C and desolvation temperature set to 550 ◦ C. All runs
were acquired with a detector voltage of 2450 V with the following
hybrid analog-to-digital converter (ADC) parameters: an ampli-
tude threshold of 2 V, an ion area threshold of 3 V and an ion area
offset of 15 V. Instrument calibration was achieved with an auto-
matic Intellistart application included in the MassLynx software
package (Waters, UK), and acquisition was conducted with a solu-
tion of 0.1% formic acid:0.1 M NaOH:acetonitrile at a ratio of 1:1:8
to achieve less than 1 ppm across 14 monoisotopic masses such
as [M + H]+ . LockSpray setup was also performed prior to acqui-
sition with leucine enkephalin (leu-enk) [M + H]+ = 556.2771, and
DRE lenses were automatically adjusted to allow for maximum
transmission with a solution at 1 ng uL−1 and an infusion rate of
5 uL min−1. An average ion area of 32 was also obtained from the
detector setup with leu-enk.
2.5. UPLC–MSE data processing with progenesis QI
Raw UPLC–MSE data files were processed and grouped by
conditions as described [11]. The identification and relative quan-
tification based on ion accounting of putative metabolites were
performed (default parameters) [14] via Progenesis QI v.2.0 (Non-
linear Dynamics, Waters, UK). The metabolites were identified “on
the fly” with the use of precursor ion exact mass, isotopologue dis-
tribution match and fragment mass ion matching with the Human
Metabolome Database (HMDB) [13] and filtered with the urine
metabolites database.
3
2.6. Data analysis
Data were filtered based on metabolite replication over indi-
vidual analytical data acquisition, and only metabolites that were
detected in all samples were considered. The remaining metabo-
lites were included in a “volcano” plot overview to evaluate which
compounds were up- and down-regulated over T3 and T5. The
metabolites were classified according to class and super class using
the Human Metabolome Database (HMDB) classification system
with an initial evaluation of the data included in the small molecule
pathway database (SMPDB) [6]. Metabolic pathways were also
investigated using the KEGG database [8].
3. Results
In the present study, we compared urine metabolomics as a
biological matrix before and after a soccer match in athletes to
investigate alterations in metabolites in response to acute inter-
mittent exercise. For these purposes, urine samples were taken
at rest as well as immediately after the soccer match. Using the
NTA approach, a total of 1091 metabolites were identified (Table
2 Supplemental Material) and classified as follows: up-regulated,
unchanged and down-regulated (Fig. 1 Supplemental Material).
Metabolites were grouped and measured according to their
appearance (%) in super classes (Fig. 2 Supplemental Material).
The biological process analysis revealed that most up-regulated
metabolites were involved with lipids and lipid-like molecules
(46%), which reflected the fatty acyls class. Organic acids and
derivatives, as well as organoheterocyclic compounds (14% and
13%, respectively) reflected the benzene and substituted deriva-
tives class. There were only a few metabolite changes (∼5%) for
nucleosides, nucleotides, and their analogs, as well as phenyl-
propanoids, polyketides, and organo-oxygen compounds. Alkaloids
and derivatives and no super classes were identified with a group
that corresponded to approximately 2% of metabolites (Fig. 2a).
Similarly, unchanged metabolites were involved with lipids and
lipid-like molecules (31%), organic acids and derivatives (17%)
and organoheterocyclic compounds (19%). Additionally, there were
only a few metabolite changes (∼5%) to nucleosides, nucleotides,
4 E. Prado et al. / International Journal of Mass Spectrometry 418 (2017) 1–5

Fig. 3. Playing soccer affects glucose (a), blood urea nitrogen (b), urate (c) and creatinine (d) metabolism after the match.

Fig. 4. Playing soccer did not affect sodium (a), potassium (b), chloride (c) or calcium (d) after the match.

and their analogs, as well as phenylpropanoids, polyketides, and
organo-oxygen compounds (Fig. 2b). Down-regulated molecules
were also not different in terms of the distribution of major
classes compared to up-regulated molecules and were mainly
involved with lipids and lipid-like molecules (27%), organic acids
and derivatives (14%) and organoheterocyclic compounds (15%).
There were only a few metabolite changes (6% to 8%) to nucleo-
sides, nucleotides, and their analogs, as well as phenylpropanoids
and polyketides, and organo-oxygen compounds (Fig. 2c).
In the urine samples, metabolites were identified as being
significantly different between up-regulated and down-regulated
molecules. Of these molecules, 188 metabolites were up-regulated,
while 338 metabolites were down-regulated. On the other hand,
565 identified and measured metabolites did not change sig-
nificantly according to the statistics used in this study. Most
up-regulated prominent molecular species were fatty acyls, car-
boxylic acids and derivatives as well as steroids and steroid
derivatives. Among the lowest expression of down-regulated
molecules were fatty acyls, carboxylic acids and their derivatives,
including benzene and substituted derivatives (Fig. 3 Supplemental
Material).
 E. Prado et al. / International Journal of Mass Spectrometry 418 (2017) 1–5
After metabolites were determined to be up- or down-regulated,
some metabolites were grouped into classes to analyze metabolic
pathways involved in purine metabolism and to study hyperam-
monemia (Fig. 2 and Table 1 Supplemental Material).
To follow-up on our findings in urine samples, we used point-of-
care equipment analysis for measuring capillary blood metabolites.
Glucose significantly increased by 35%, whereas urate increased by
16%, and uremia increased by 17% without changes in creatine-
mia (Fig. 3). Kalemia decreased by ∼15% in response to exercise,
although sodium, calcium and chloride blood concentrations did
not change (Fig. 4).
4. Discussion
We measured urinary metabolites in semi-professional soc-
cer players before and after a soccer match using NTMS analysis.
Several metabolic pathways were modified by exercise during a
soccer match, and the use of a non-invasive metabolome analysis in
future studies could offer important information for understanding
human metabolism and could also be used as a tool to help athletes
achieve better performance.
In our experiments, different urinary metabolites from diverse
metabolic pathways were significantly changed in response to
exercise. Due to our interest in energy metabolism, we chose to
follow-up on an important bioenergetics pathway. During high-
intensity exercise, the decrease in the ATP/ADP ratio activates
myokinase activity. This enzyme phosphorylates an ADP using
another ADP as both the phosphate and energy donor. The reaction
products are ATP, which can be used as an energy source, and AMP.
AMP is deaminated to IMP by AMP deaminase, producing hypox-
anthine, and is finally excreted as urate [12]. Due to the need for
AMP deamination, other intermediate metabolites such as hypox-
anthine and IMP increase in muscle after high-intensity exercise
[4,7].
Following AMP metabolism, ammonia and urate are produced
with similar stoichiometries. Thus, it is feasible to estimate the use
of ADP to regenerate ATP. Additionally, it has been shown that
ammonia is highly toxic and is related to fatigue during exercise
[2]. Exercise-induced hyperammonemia occurs via both amino acid
catabolism and AMP deamination, and that activity increases dur-
ing vigorous exercise as a result of a decreased energy state [9,12].
In the present study, we showed that hypoxanthine and related
metabolites were up-regulated in urine after the soccer match,
which suggested that AMP deamination also increased. We also
showed that glucose increased in blood as both urea and urate in
blood. These data together can show that during the soccer match,
there was an increase in ATP use through ADP synthesis via myoki-
nase [15]. These data were reinforced by the measurements of both
urea (ammonia final metabolite) and urate (IMP final metabolite)
in blood and through the increase in several urate metabolic inter-
mediates in urine. Carbohydrate availability (carbohydrates were
given to the athletes before the game and during the interval),
measured by way of glucose increase in blood, probably decreased
amino acid catabolism, which was reinforced by the small increase
in blood urea [15].
5
5. Conclusion
Our data may show that the use of urinary metabolomics can
be a less invasive way to measure metabolism in athletes during
exercise. It was demonstrated through different approaches that
NTMS may be useful for future studies that aim to design holistic
interventions to help guide training, diet, or other interventions to
improve athletic performance.
Financial support
The authors acknowledge the Abaxis Global Diagnostics; Comitê
Olimpíco do Brasil (BOC); Conselho Nacional de Desenvolvimento
Científico e Tecnológico (CNPq); Coordenação de Aperfeiçoamento
de Pessoal de Nível Superior (CAPES); Financiadora de Estudos
e Projetos (FINEP); Fundação Carlos Chagas Filho de Amparo à
Pesquisa do Estado do Rio de Janeiro (FAPERJ); Merck-Sigma-
Aldrich; SelectScience; Universidade Federal do Estado do Rio de
Janeiro (UNIRIO) and Waters Corporation for the project funding.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.ijms.2017.02.002.
References
[1] P.J. Atherton, et al., Exercise and regulation of protein metabolism, Prog. Mol.
Biol. Transl. Sci. 135 (2015) 75–98.
[2] E.W. Banister, B.J. Cameron, Exercise-induced hyperammonemia: peripheral
and central effects, Int. J. Sports Med. 11 (2) (1990) S129–S142.
[3] A. Bassini, L.C. Cameron, Sportomics: building a new concept in metabolic
studies and exercise science, Biochem. Biophys. Res. Commun. 445 (4) (2014)
708–716.
[4] S. Broberg, K. Sahlin, Adenine nucleotide degradation in human skeletal
muscle during prolonged exercise, J. Appl. Physiol. (1985) 67 (1) (1989)
116–122.
[5] X. Feng, et al., Mass spectrometry in systems biology: an overview, Mass
Spectrom. Rev. 27 (6) (2008) 635–660.
[6] A. Frolkis, et al., SMPDB: the small molecule pathway database, Nucleic Acids
Res. 38 (2010) D480–D487 (Database issue).
[7] Y. Hellsten, et al., AMP deamination and purine exchange in human skeletal
muscle during and after intense exercise, J. Physiol. 520 (3) (1999) 909–920.
[8] M. Kanehisa, et al., KEGG as a reference resource for gene and protein
annotation, Nucleic Acids Res 44 (D1) (2016) D457–D462.
[9] J.P. Manfredi, E.W. Holmes, Control of the purine nucleotide cycle in extracts
of rat skeletal muscle: effects of energy state and concentrations of cycle
intermediates, Arch. Biochem. Biophys. 233 (2) (1984) 515–529.
[10] S.G. Ra, et al., Metabolomics of salivary fatigue markers in soccer players after
consecutive games, Appl. Physiol. Nutr. Metab. 39 (10) (2014) 1120–1126.
[11] M. Sedic, et al., Label-free mass spectrometric profiling of urinary proteins
and metabolites from paediatric idiopathic nephrotic syndrome, Biochem.
Biophys. Res. Commun. 452 (1) (2014) 21–26.
[12] D.J. Wilkinson, et al., Ammonia metabolism, the brain and fatigue; revisiting
the link, Prog. Neurobiol. 91 (3) (2010) 200–219.
[13] D.S. Wishart, et al., HMDB 3. 0 – the human metabolome database in 2013,
Nucleic Acids Res. 41 (2013) D801–D807, Database issue.
[14] A. Zhang, et al., Rapidly improved determination of metabolites from
biological data sets using the high-efficient TransOmics tool, Mol Biosyst 10
(8) (2014) 2160–2165.
[15] J. Zielinski, K. Kusy, Hypoxanthine: a universal metabolic indicator of training
status in competitive sports, Exerc. Sport Sci. Rev. 43 (4) (2015) 214–221.