Isotope-labeled differential profiling of metabolites using N-benzoyloxysuccinimide derivatization coupled to liquid chromatography/high-resolution tandem mass spectrometry

Article abstract of DOI:10.1002/rcm.7264

Rationale An isotopic labeling strategy based on derivatizing amine-containing metabolites has been developed using light (¹²C₆) and heavy (¹³C₆) N-benzoyloxysuccinimide reagents for semi-targeted metabolomic applications. Methods Differentially labeled samples were combined and analyzed simultaneously by liquid chromatography/high-resolution tandem mass spectrometry (LC/HR-MS/MS) to compare relative amounts of amine-containing metabolites. The selectivity of the reaction was determined with model metabolites and was shown to also be applicable to thiol and phenol moieties. The potential for relative quantitation was evaluated in cell extracts and the method was then applied to quantify metabolic perturbations occurring in human cultured cells under normal vs. oxidative stress conditions. Results A total of 279 derivatized features were detected in HL60 cell extracts, 77 of which yielded significant concentration changes upon oxidative stress treatment. Based on accurate mass measurements and MS/MS spectral matching with reference standard solutions, 10 metabolites were clearly identified. Derivatized compounds were found to have diagnostic fragment ions from the reagent itself, as well as structurally informative ions useful for metabolite identification. Conclusions This simple derivatization reaction can be applied to the relative quantitation of amine-, thiol- and phenol-containing compounds, with improved sensitivity and chromatographic peak shapes due to the increased hydrophobicity of polar metabolites not readily amenable to reversed-phase LC/MS analysis.

Full text of DOI:10.1002/rcm.7264

Research Article
Received: 23 April 2015
Revised: 15 June 2015
Accepted: 21 June 2015
Published online in Wiley Online Library
Rapid Commun. Mass Spectrom. 2015, 29, 1632–1640
(wileyonlinelibrary.com) DOI: 10.1002/rcm.7264
Isotope-labeled differential profiling of metabolites using
N-benzoyloxysuccinimide derivatization coupled to liquid
chromatography/high-resolution tandem mass spectrometry
Michel Wagner, Leanne B. Ohlund, Tze Chieh Shiao, Amélie Vézina, Borhane Annabi,
*
René Roy and Lekha Sleno
Université du Québec à Montréal, Chemistry Department/Pharmaqam, Montréal, QC, Canada
RATIONALE: An isotopic labeling strategy based on derivatizing amine-containing metabolites has been developed
using light (12C6) and heavy (13C6) N-benzoyloxysuccinimide reagents for semi-targeted metabolomic applications.
METHODS: Differentially labeled samples were combined and analyzed simultaneously by liquid chromatography/high-
resolution tandem mass spectrometry (LC/HR-MS/MS) to compare relative amounts of amine-containing metabolites. The
selectivity of the reaction was determined with model metabolites and was shown to also be applicable to thiol and phenol
moieties. The potential for relative quantitation was evaluated in cell extracts and the method was then applied to quantify
metabolic perturbations occurring in human cultured cells under normal vs. oxidative stress conditions.
RESULTS: A total of 279 derivatized features were detected in HL60 cell extracts, 77 of which yielded significant
concentration changes upon oxidative stress treatment. Based on accurate mass measurements and MS/MS spectral
matching with reference standard solutions, 10 metabolites were clearly identified. Derivatized compounds were found
to have diagnostic fragment ions from the reagent itself, as well as structurally informative ions useful for metabolite
identification.
CONCLUSIONS: This simple derivatization reaction can be applied to the relative quantitation of amine-, thiol- and
phenol-containing compounds, with improved sensitivity and chromatographic peak shapes due to the increased
hydrophobicity of polar metabolites not readily amenable to reversed-phase LC/MS analysis. Copyright © 2015 John
Wiley & Sons, Ltd.
The goal of metabolomic studies is to analyze low molecular
weight endogenous metabolites present in a biological
sample and track concentration changes across different
samples.[1,2] In comparison with other ’omics’ approaches,
these analytes are arguably more chemically diverse, and
there is currently no universal platform capable of measuring
stable isotope labeled internal standards (IS) can be spiked
into samples, with the advantage of having the same
ionization efficiency and chromatographic elution behavior
as their non-labeled counterparts. Nevertheless, isotopically
labeled compounds are not available for every analyte. It is
possible to use a limited subset of IS for each chemical
class,[8] but this solution is not ideal since ionization
efficiencies and matrix effects are not comparable for all
analytes.[9]
all metabolites in
a
sample.[3] Liquid chromatography
coupled to mass spectrometry (LC/MS) is one of the most
popular techniques for this type of analysis.[4] Features,
i.e., signals characterized by a given m/z-RT (retention time)
pair, exhibiting significant changes are typically filtered using
statistical tools such as multivariate analysis (e.g., principal
component analysis and partial least-squares discriminant
analysis).[5] Once statistically interesting features are found,
the chemical structure of the corresponding compounds
should be elucidated.[6,7]
Quantitation of these metabolites is often also desirable;
however, this is not straightforward. Absolute quantitation
relies on a reference standard solution linking the analyte
signal to a corresponding concentration. For this purpose,
Another option uses relative quantitation to determine
metabolite changes between two samples (i.e., differential
analysis). This concept relies on sample derivatization with a
reagent that is available in two isotopic forms: one (heavy) form
(or SIL analog) and one light (unlabeled) form. Several isotope-
labeling strategies have been developed to derivatize specific
chemical groups, including amines, carboxylic acids, and
aldehydes and ketones.[10,11] Amine-containing metabolites
have very diverse chemical structures (e.g., amino acids,
polyamines, catecholamines, nucleotides) and play essential
biochemical roles. Reagents used for amine isotope labeling
fall into several classes: aldehydes,[12,13] dansyl chloride
and derivatives,[14,15] pentafluorophenyl-activated esters of
poly(ethylene glycol),[16] methyl acetimidate,[17] and
* Correspondence to: L. Sleno, Chemistry Department,
Université du Québec à Montréal, PO Box 8888 Downtown
Station, Montréal, Québec, Canada, H3C 3P8.
E-mail: sleno.lekha@uqam.ca
N-hydroxysuccinimide
derivatives.[18–21]
Nevertheless,
most of these approaches have been investigated for
targeted analysis. Metabolomic applications has been
Rapid Commun. Mass Spectrom. 2015, 29, 1632–1640
Copyright © 2015 John Wiley & Sons, Ltd.

Differential analysis of amine metabolites in cells
demonstrated with dansyl or dansyl-like derivatives of
amine metabolites, using a 13C2 analog as the heavy
reagent.[14,15,22,23]
under a humidified atmosphere containing 5% CO2. The
stock suspension was diluted to a concentration of 5 × 106
cells per mL.
In this work, an isotope-labeling strategy based on an N-
hydroxysuccinimide ester reagent with a heavy reagent
incorporating six 13C atoms into a benzyl ring has been
developed. Following the evaluation of this approach using
a set of standard metabolites and cell extracts, the method
was applied to a cellular model of oxidative stress to
demonstrate its potential for differential analysis.
Differential analysis/oxidative stress experiments
Aliquots (0.4 mL) of HL60 cells (5 × 106 cells/mL) (n = 5) were
treated with either 20 μL of water (control), or 20 μL of a 20 mM
H2O2 solution (treated), for 20 min at 37°C with gentle mixing.
Metabolites were extracted (see below) and cell extracts were
evaporated to dryness. Control series were derivatized with
light and heavy reagents, whereas treated samples were
derivatized with light reagent only. Derivatized extracts were
mixed 1:1 (light/heavy), resulting in five control:control and
five treated:control samples. Quality control samples were
prepared by pooling equal aliquots of control samples.
EXPERIMENTAL
Materials
Benzoic acid, (ring-13C6)-benzoic acid, formic acid (>98%), 2-
aminoisobutyric acid (AIBA), N-acetylcysteine (NAC),
Metabolite extraction
adenosine
(Ado),
adenosine
5’-diphosphate
(ADP),
Cell culture samples were centrifuged (3500 rpm, 5 min at 4°C)
and supernatants were discarded. Pellets were then washed
twice with 1 mL cold phosphate-buffered saline (PBS). To the
resulting pellets, 300 μL MeOH/H2O 1:1 (v/v) was added
and samples were sonicated (8 min). Ice-cold MeOH (300 μL)
was then added, samples were centrifuged (14000 rpm, 8 min
at 4°C), and 550 μL of supernatant was retrieved into a new
tube. An additional 200 μL of MeOH/H2O 1:1 (v/v) was added
to the pellet, and 200 μL of resulting supernatant were
combined with the previous extract (total volume of 750 μL).
Extracts were evaporated to dryness prior to derivatization.
adenosine cyclic 2’,3’-monophosphate (AMPc), alanine (Ala),
arginine (Arg), asparagine (Asn), aspartic acid (Asp),
citrulline (Cit), cystathionine, cysteine (Cys), γ-aminobutyric acid
(GABA), glutamic acid (Glu), glutamine (Gln), glutathione
(GSH), glutathione disulfide (GSSG), glycine (Gly), histidine
(His), hydroxykynurenine (Kyn-OH), hydroxyproline (Pro-OH),
isoleucine (Ile), leucine (Leu), lysine (Lys), methionine (Met),
ornithine (Orn), phenylalanine (Phe), phosphorylethanolamine,
proline (Pro), serine (Ser), taurine (Tau), threonine (Thr),
tryptamine (Trpmn), tryptophan (Trp), tyrosine (Tyr) and valine
(Val) were obtained from Sigma-Aldrich (Oakville, ON, Canada).
N-Hydroxysuccinimide was purchased from Alfa Aesar (Ward
Hill, MA). N,N’-Dicyclohexylcarbodiimide, ethyl acetate and
hexane were purchased from Acros Organics-Fisher Scientific
(Ottawa, ON, Canada). Acetonitrile (ACN), monopotassium
phosphate and hydrogen peroxide (30%) were from Caledon
Laboratory Chemicals (Georgetown, ON, Canada). Methanol
was obtained from EMD (Gibbstown, NJ, USA) and ultrapure
water was supplied by a Synergy® UV purification system
(Millipore, Billerica, MA, USA). Disodium tetraborate
decahydrate was obtained from Alfa Aesar (Ward Hill, MA,
USA). Potassium chloride, sodium chloride and anhydrous
dipotassium phosphate were from Anachemia (Rouses Point,
NY, USA). The HL-60 promyelocytic human cell line was
acquired from the American Type Culture Collection
(Manassas, VA, USA). Iscove’s modified Dulbecco’s medium
and Trypan blue solution (0.4%) were from Gibco (Life
Technologies, Burlington, ON, Canada). Fetal bovine serum
(FBS) was from HyClone (Thermo Scientific, Logan, UT,
USA). Penicillin-streptomycin glutamine solution was
purchased from Wisent (St-Bruno, QC, Canada).
Relative quantitation experiments
Five extracts were created each using four aliquots of 1 mL
HL60 cell suspensions (5 × 106 cells/mL), subsequently
divided into multiple aliquots representing extracts from 1, 2
or 5 million cells each. Light and heavy labeled samples were
mixed in 1:1 proportions to prepare five series of the following
mixtures (light:heavy ratios): 1:1, 1:2, 1:5, 2:1, 2:2, 5:1, 5:5.
Derivatization reaction
Dried samples were reconstituted in 120 μL of derivatization
solvent (50 mM borate buffer containing 5% ACN, pH 9),
followed by the addition of 20 μL N-benzoyloxysuccinimide
(light or heavy, 20 mM in ACN), and incubated for 30 min
at 25°C, while mixing (Fig. 1).
LC/MS analysis
Reversed-phase liquid chromatography was performed on a
Shimadzu Nexera ultra-high-performance liquid chroma-
tography (UHPLC) system with mobile phases A (H2O) and
B (ACN), each containing 0.1% formic acid. Separation was
achieved using a Cogent C18 column (150 × 2 mm, 4 μm),
with a Phenomenex SecurityGuard C18 (4 × 2.0 mm) guard
column, at 40°C, using a gradient elution as follows: 5% B
(0–2 min), 25% (6 min); 65% (17 min); 90% (18–20 min), at a
flow rate of 0.4 mL/min.
Derivatization reagents (N-benzoyloxysuccinimide and N-
benzoyl-ring-13C6-oxy-succinimide) were synthesized in-
house (method described in the Supporting Information).
Two standard solutions of model metabolites (400 μM each)
were prepared by mixing individual stock solutions into
two standard mixes (A and B).
The LC system was coupled to a hybrid quadrupole-time-
of-flight mass spectrometer (AB Sciex Triple TOF® 5600,
Concord, ON, Canada) using a DuoSpray™ ion source
operated in positive electrospray mode. Source conditions
were: source voltage 5000 V, curtain gas 35 psi, temperature
Cell culture
HL-60 cells were cultured in Iscove’s modified Dulbecco’s
medium containing 20% (v/v) FBS, 2 mM glutamine,
100 units/mL penicillin, 100 μg/mL streptomycin, at 37°C
Rapid Commun. Mass Spectrom. 2015, 29, 1632–1640
Copyright © 2015 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/rcm

M. Wagner et al.
filtering criteria were: peak width ≥2.5 s, intensity ≥100 counts
per second (cps), and sample-to-control intensity ratio ≥3, m/z
tolerance of 25 ppm, and XIC width of 25 mDa. The differential
isotope filtering step was performed to find peak pairs with m/z
differences (Δm/z) of 6.0201 and 12.0402, for singly and doubly
labeled compounds, respectively, using a set of control cell
extracts derivatized in a 1:1 ratio. This peak list was then
exported into Microsoft Excel 2010 and further annotated and
filtered to remove adducts and multiply charged species or
multimers. Accurate peak integration of remaining features
was performed using MultiQuant 2.1, and average peak area
ratios (light/heavy derivative) were compared using a two-
sided t-test. All calculations were performed in Excel, including
t-statistics, adjusted degrees of freedom and resulting p-value.
For structural elucidation of metabolites, underivatized
accurate masses were calculated and subjected to database
searching (METLIN[24,25] and HMDB[26,27]). Metabolite
identification was achieved by retention time (RT) and
HRMS/MS spectral matching, with derivatized commercially
available reference standards.
Figure 1. Derivatization reaction using benzoyloxysuccinide
reagent, with example shown with free amine group being
labeled.
500°C, and nebulizer and drying gases (GS1/2) both 50 psi.
The declustering potential was 60 V and MS/MS experiments
used collision-offset voltages of 30 10 V. TOF-MS spectra
(m/z 110–1000,250 ms accumulation time) were acquired,
followed by MS/MS experiments (m/z 50–860, 100 ms
accumulation time), using information-dependent acquisition
and dynamic background subtraction, with the five most
intense precursor ions. Follow-up experiments for metabolite
identification used an inclusion list of selected precursor ions,
with an MS/MS range of m/z 50–625 and 200 ms accumulation
time, for increased sensitivity and higher quality MS/MS data.
Mass calibration was performed automatically every four
injections, using a mix of standards from m/z 121 to 922 for
TOF-MS, and, in MS/MS mode, using product ions of
reserpine (MH+) ranging from m/z 74 to 609.
RESULTS AND DISCUSSION
Derivatization reaction
The derivatization reagent was designed to target amine
groups (Fig. 1), since succinimidyl esters are activated
carboxylic acids commonly used to form amide bonds.[28]
They are highly reactive electrophiles towards primary and
secondary amines (nucleophilic substitution). To be applicable
for differential analysis, light and heavy versions of the reagent
were synthesized, the heavy one containing six 13C atoms on
the aromatic ring. In the present study, the 6 Da difference
Data processing and statistical analysis
Raw LC/MS/MS data was processed using Metabolite Pilot™
(version 1.5, AB Sciex) for peak picking and peak-pair finding,
using a custom differential isotope-filtering tool. LC/MS
Figure 2. Extracted ion chromatograms of model metabolites in standard solutions containing
8.5 μM of each metabolite: (a) standard mixture A, derivatized; (b) standard mixture B,
derivatized; (c) standard mixture A, underivatized; (d) standard mixture B, underivatized.
*: doubly derivatized metabolite. **: triply derivatized metabolite.
wileyonlinelibrary.com/journal/rcm
Copyright © 2015 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2015, 29, 1632–1640

Differential analysis of amine metabolites in cells
was chosen so as to introduce a highly unique isotope pattern
for filtering purposes when analyzing complex samples.
The other aspect considered in the design of the heavy
reagent was the nature of the isotope introduced. Given the
structure of organic compounds, labels usually include 2H
or 13C. While the use of the former is more widespread and
typically cheaper, it also has the potential drawback of
extensive isotope effects in RP-LC.[29] Typically, deuterated
analogs have weaker hydrophobic interactions than their
non-deuterated counterparts, and therefore exhibit earlier
retention times in RP-LC.[29,30] Therefore, complete co-elution
of light and heavy derivatives is rarely possible.[12,20,31] This
lack of co-elution would be detrimental to both relative
quantitation and metabolite peak finding. In this study,
peak pairs (light/heavy derivatives) were systematically
inspected for proper co-elution and no isotope effects were
observed.
A set of standard metabolites was used to characterize the
selectivity of the reaction. Two standard mixes (A and B) were
prepared to alleviate possible isobaric interferences. Derivatives
of amine-containing metabolites were obtained, and the
presence of multiple amine groups led to the formation of
multiply labeled metabolites. The only notable exception was
the case of aromatic amines, which did not react with the
reagent, given that the nitrogen lone-pair of such amines is
delocalized into the aromatic system (adenine moiety). Also
amide groups do not react, since they exhibit more acidic
behavior than their amine counterparts, therefore are much less
reactive. On the other hand, the reaction was also found to occur
with free thiols (Cys, NAC, GSH) and phenols (tyrosine,
hydroxykynurenine).
RP-LC of endogenous metabolites can be quite challenging,
because they are often very polar small molecules. The
limitation of RP-LC is illustrated in Fig. 2 by the extracted ion
chromatograms of a set of underivatized model metabolites:
31 of the 33 metabolites eluted earlier than 1.5 min, and would
therefore have considerable matrix effects in biological samples.
Moreover, elution of a large number of compounds in a short
time window would impair peak finding algorithms, bias
quantitation (presence of multiple interferences), limit MS/MS
triggering in IDA experiments as well as the use of retention
time matching. The benzyl group in the derivatization reagent
increases chromatographic retention and therefore improves
metabolite separation. The derivatized metabolites also exhibit
much higher signals than their underivatized counterparts.
Figure 3. TOF-MS spectra of Leu illustrating typical redundant
species observed.
Rapid Commun. Mass Spectrom. 2015, 29, 1632–1640
Copyright © 2015 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/rcm

M. Wagner et al.
For the 31 derivatized model metabolites, the median of peak
intensity ratio (derivatized/underivatized) was 9.1. This ratio
was >30 for over 25% of them, with a maximal value close to
~400 for Cys (doubly derivatized).
metabolite is shown for leucine in Fig. 3. In this case, the
pair at m/z 236.1287 and m/z 242.1489 would be retained
in the filtered list but minor adduct and cluster ions
can easily be removed prior to differential analysis of
metabolites.
Peak picking and pair finding
Relative quantitation
A generic peak-picking step yielded an initial list of LC/MS
peaks. This list can then be filtered to find peak pairs
having the same RT and 6.020 or 12.040 Da differences for
singly and doubly derivatized compounds, respectively.
The resulting list was then annotated to remove redundant
features from cluster or adduct ions. An example of the
multiple ion species seen in the mass spectrum for a given
Cell extracts were employed to evaluate that signal ratios
between heavy and light derivatives could effectively track
concentration ratios. Other research groups have investigated
the quantitative performance of their strategies with
standard solutions of model metabolites (i.e., non-biological
samples).[12,14,15,17,20] The experimental design relied on
mixing known amounts of derivatized metabolites (light and
heavy) in various ratios to generate a calibration curve. This,
however, does not directly ascertain the ability of the
derivatization reaction to track quantitative changes reliably
in specific biological samples. The experimental design chosen
to test the present approach employed five independent
pooled cell extracts, each being divided into different aliquots.
A
representative sample was used to identify singly
derivatized features, leading to a final list of 115 derivatized
peak pairs (after removing redundant features, including
adducts and in-source fragments). Average peak area ratios
(light/heavy) were determined and coefficients of variations
(CVs) across the five sets were calculated to estimate the
reproducibility of the method, accounting for sample
variability related to extraction, derivatization and LC/MS
analysis. The median CV was 15.3%, with a large proportion
(75%) having CVs less than 25%.
The agreement between experimental measurements (mean
peak area ratios) and expected values (known concentration
ratios) was estimated using several parameters. First, coefficients
of determination (r2) measured how well values correlate, of
which 92% of features exhibited r2 ≥0.95. The slopes of the
corresponding linear regressions were also compared (ideal
slope of 1), and 70% yielded a slope between 0.7 and 1.3. Lastly,
the best estimation of accuracy is arguably the bias (i.e., relative
difference) between known concentration ratios and mean peak
area ratios, where in the case of the 115 peak pairs, the median
bias was –5.7%. Representative results for 10 model metabolites
detected in cell extracts are presented in Table 1, illustrating the
individual calculated peak area ratios from 1:1, 2:2 and 5:5
extract ratios and their associated %CVs, as well as linearities
(slope and r2) from 1:5 to 5:1 extracts ratios.
Figure 4. Experimental design of the differential analysis for
oxidative stress study in cells.
Table 2. Identified metabolites with significant changes following H2O2 treatment of HL60 cells
Metabolite
Formula (underivatized)
# Labels
Fold change
p-value
1
2
3
4
5
6
7
8
9
10
Taurine
C2H7NO3S
C5H9NO4
C7H14N2O4S
C10H17N3O6S
C4H9NO2
C2H8NO4P
C5H11NO2S
C2H4NO2
C9H11NO3
C5H11NO3S
1
1
2
2
1
1
1
1
2
1
À2.6
À2.1
À2.0
À1.9
À1.8
À1.6
À1.6
À1.6
À1.5
4.0
2.48e-06
1.06e-05
1.00e-05
2.72e-02
3.26e-03
1.72e-04
3.81e-03
2.72e-04
5.01e-03
2.61e-04
Glutamic acid
Cystathionine
Glutathione (GSH)
γ-Aminobutyric acid (GABA)
Phosphorylethanolamine
Methionine
Glycine
Tyrosine
Methionine sulfoxide
wileyonlinelibrary.com/journal/rcm
Copyright © 2015 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2015, 29, 1632–1640

Differential analysis of amine metabolites in cells
Differential analysis
analyzed at the beginning, in the middle and at the end of
the sequence and showed no significant deviations in RT over
the course of the analytical run.
The present method was applied to evaluate its ability to
track metabolic perturbations in biological samples. HL60
cells were treated with H2O2 to mimic oxidative stress. This
metabolic perturbation was employed to test the
derivatization approach, not necessarily to accurately model
a given pathological process. The experimental design is
illustrated in Fig. 4. The amount of cells per sample (2 × 106)
was chosen in the range previously investigated during the
quantitative evaluation of the method. A set of quality control
(QC) samples (5 pooled derivatized controls extracts) was
The pooled QC sample was also used for data processing in
order to identify feature pairs. The search was performed for
singly and doubly derivatized species, and resulted in a raw list
of 398 feature pairs. This list was inspected and redundant
features were removed, including sodium and potassium
adducts, dimers, trimers, and doubly charged species. Dimers,
trimers and doubly charged species were identified by
manually inspecting MS spectra and were only removed if the
monomeric feature was also identified. Interestingly, dimers
Figure 5. Metabolites, identified by retention time and MS/MS spectral matching
with authentic standards, found to change in HL60 cells upon H2O2 treatment. Site
of benzyl label derivatization indicated by *.
Table 3. LC/MS/MS data (retention times and fragment ions) for identified metabolites changing in H2O2-treated HL60 cells
Precursor m/z
12C 13C
Diagnostic ions
RT
(min)
Metabolite
51/55 77/83 105/111
Fragment ions
134.0604, 162.0614
72.0478*, 86.0613*, 90.0572*, 149.0657, 191.0790
148.0755
148.0749, 149.0780
102.0563*, 130.0508*, 160.0748, 188.0695, 206.0823,
234.0767
56.0532*, 61.0140*, 104.0538*, 130.0660, 160.0761,
206.0815, 208.0795, 236.0737
56.0527*, 104.0503, 131.0155*, 130.0654, 149.0259*,
160.0755, 206.081, 252.0685
Glycine
γ-Aminobutyric acid
Taurine
5.9 180.0655 186.0856
7.1 208.0968 214.1169
4.3 230.0481 236.0682
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
Phosphorylethanolamine 3.1 246.0526 252.0727
Glutamic acid
6.3 252.0866 258.1067
9.5 254.0845 260.1046
6.0 270.0794 276.0995
Methionine
+
+
+
+
+
Methionine sulfoxide
Tyrosine
Cystathionine
14.4 390.1336 402.1738
9.7 431.1271 443.1673
+
+
+
+
+
119.0494*, 240.1020, 344.1286, 372.1237
114.0384*, 142.0326*, 146.0608, 160.0698, 174.0554,
192.0653, 206.0814, 218.0636, 220.0429, 226.0537,
238.0538, 240.0689, 246.0587, 263.0851, 264.0691,
281.0961, 309.0912, 385.1220, 413.1168, 431.1275
142.0322*, 170.0270*, 180.0475, 199.0534*, 216.0653,
227.0485*, 234.0762, 245.0603*, 265.0648, 283.0753,
291.0803, 303.0981, 319.0752, 360.1203, 376.0974,
394.1075, 395.1055, 413.1170, 441.1119, 516.1451
Glutathione
9.9 516.1435 528.1837
+
+
*Denotes fragments without benzyl label. Top three most intense fragments shown in bold.
Rapid Commun. Mass Spectrom. 2015, 29, 1632–1640
Copyright © 2015 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/rcm

M. Wagner et al.
Structural elucidation and metabolite identification
and trimers exhibit a characteristic isotopic distribution based
on the combination probabilities of light (m) and heavy (M)
derivatives (monomers). For instance, dimers give rise to three
peaks distanced by 6.020 Da with intensity of ratios 1:2:1,
corresponding to: [m+m+H]+ (first peak), [M+m+H]+ and
[m+M+H]+ (second peak), [M+M+H]+ (third peak). Similarly,
dimers and trimers could be observed as sodium and potassium
adducts. After removing these features, the resulting list
consisted of 279 feature pairs, used to calculate peak area ratios
in all samples.
The mean of peak area ratios for each feature was
compared between the two sets of samples (control/control
and treated/control, n = 5) using a two-sided t-test. A feature
was qualified as discriminating if the p-value associated to the
t-test was <0.05 and if the control/control mean was
changing by at least 50%. A final list of 77 statistically
significant features fell within these criteria.
Once features of interest were noted as statistically relevant,
the next step was to identify metabolites, which usually
represents the most challenging part in metabolomic
workflows. For the 77 discriminating features found, the
monoisotopic mass of the native (neutral, underivatized)
metabolite was first determined by subtracting the mass of
the label moiety, namely 105.03403 Da (C7H5O) for singly
and 209.06025 (C14H9O2) Da for doubly derivatized
metabolites. The resulting mass was entered in METLIN
and HMBD databases to search for possible chemical
structures of known metabolites within a mass tolerance of
5 ppm, which yielded a list of 38 putative metabolites. The
high resolution (~25,000) and proper mass calibration of the
TOF-MS setup enabled accurate mass measurements, crucial
for structural elucidation of unknown metabolites.
Nevertheless, metabolite identification based solely on
accurate mass is relatively limited, since it simply yields a
putative list of metabolites. The second key element used
for structural elucidation and to refine metabolite database
results was the MS/MS spectra. Moreover, MS/MS data were
used to confirm that an ion corresponded effectively to a
derivative, based on diagnostic fragment ions found to be
characteristic of the label moiety and having a 6.0201 Da
difference between the light and heavy labels, including
cyclobutadienyl cation (C4H3+, m/z 51.0229 / 13C4H3+, m/z
55.0363), phenylium ion (C6H5+, m/z 77.0391/13C6H5+, m/z
83.0593), benzoyl cation (C7H5O+, m/z 105.0340/13C6CH5O+,
m/z 111.0542).
Finally, when available, metabolites changing significantly
with H2O2 treatment were identified by comparison with
commercially available standards. Ten discriminating
metabolites were clearly identified based on RT and MS/MS
matching with standard derivatized metabolites. There was
a significant decrease in taurine, glutamic acid, cystathionine,
GSH, GABA, phosphorylethanolamine, methionine, glycine,
and tyrosine, as well as a significant increase in methionine
Figure 6. Overlaid extracted ion chromatograms for the ten
derivatized metabolites found to be significantly changed
after H2O2 treatment, from a representative sample with
treated (light, in red) and control (heavy, in blue) cells. 2×
indicates double-derivatization and
* denotes that the
glutamic acid peak has been reduced 4-fold for clarity.
Figure 7. High-resolution MS/MS spectra of light and heavy derivatized methionine (a) and
methionine sulfoxide (b) from a representative cell extract sample. Proposed fragmentation
pathways are shown on the structure of derivatized methionine and a few specific product
ions are drawn within the MS/MS spectrum of light-derivatized Met.
wileyonlinelibrary.com/journal/rcm
Copyright © 2015 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2015, 29, 1632–1640

Differential analysis of amine metabolites in cells
sulfoxide. Table 2 lists the resulting identified metabolites
found to be significantly changed in the oxidative stress cell
model, along with corresponding fold changes and p-values.
The structures of these identified metabolites are shown in
Fig. 5, illustrating the primary amine, thiol and phenol label
sites seen for these compounds. LC/MS/MS data, including
retention times, exact masses (m/z) of derivatized metabolites,
diagnostic (label-related) as well as structurally informative
fragment ions, are listed in Table 3. Figure 6 shows the
overlaid light- and heavy-labeled metabolite extracted ion
chromatograms for all ten identified species found to be
Acknowledgements
We would like to thank Suma Rumagiri and Eva Duchoslav
from AB Sciex for the development of the custom
differential isotope-filtering tool in Metabolite Pilot 1.5
software. We would also like to acknowledge funding of
this project by Pharmaqam and FRQNT. R.R. also
acknowledges support from NSERC for a Canada Research
Chair in Therapeutic Chemistry. B.A. holds an Institutional
Research Chair in Cancer Prevention and Treatment at
UQÀM.
significantly changed after H2O2 treatment, from
a
representative sample with treated (light) and control (heavy)
cells. All derivatized metabolites yielded at least two of the
diagnostic ion pairs for benzyl-labeled species as well as
some structurally informative ions. MS/MS matching was
performed based on accurate mass ( 10 ppm) as well
REFERENCES
[1] O. Fiehn. Metabolomics – the link between genotypes and
phenotypes. Plant Mol. Biol. 2002, 48, 155.
[2] J. K. Nicholson, I. D. Wilson. Understanding ’Global’
Systems Biology: Metabonomics and the continuum of
metabolism. Nat. Rev. Drug Discov. 2003, 2, 668.
[3] A. Zhang, H. Sun, P. Wang, Y. Han, X. Wang. Modern
analytical techniques in metabolomics analysis. Analyst
2012, 137, 293–300.
[4] H. G. Gika, G. A. Theodoridis, R. S. Plumb, I. D. Wilson.
Current practice of liquid chromatography-mass spectrometry
in metabolomics and metabonomics. J. Pharm. Biomed. Anal.
2014, 87, 12.
as relative intensities between fragment ions. As
a
representative example, high-resolution MS/MS data is
shown for methionine (Met) and methionine sulfoxide (Met
(O)) in Fig. 7, along with the proposed fragmentation
reactions from both label-specific ions as well as structurally
informative ions.
[5] J. Trygg, E. Holmes, T. Lundstedt. Chemometrics in
metabonomics. J. Proteome Res. 2007, 6, 469.
CONCLUSIONS
[6] W. Dunn, A. Erban, R. M. Weber, D. Creek, M. Brown,
R. Breitling, T. Hankemeier, R. Goodacre, S. Neumann,
J. Kopka, M. Viant. Mass appeal: metabolite identification
in mass spectrometry-focused untargeted metabolomics.
Metabolomics 2013, 9, 44.
[7] K. Scheubert, F. Hufsky, S. Bocker. Computational mass
spectrometry for small molecules. J Cheminformatics 2013,
5, 12.
[8] R. L. Shaner, J. C. Allegood, H. Park, E. Wang, S. Kelly,
C. A. Haynes, M. C. Sullards, A. H. Merrill. Quantitative
analysis of sphingolipids for lipidomics using triple
quadrupole and quadrupole linear ion trap mass
spectrometers. J. Lipid Res. 2009, 50, 1692.
[9] D. H. Chace, T. Lim, C. R. Hansen, B. W. Adam, W. H.
Hannon. Quantification of malonylcarnitine in dried
blood spots by use of MS/MS varies by stable isotope
internal standard composition. Clin. Chim. Acta 2009,
402, 14.
[10] P. Bruheim, H. F. N. Kvitvang, S. G. Villas-Boas. Stable
isotope coded derivatizing reagents as internal standards
in metabolite profiling. J. Chromatogr. A 2013, 1296, 196.
[11] T. Toyo’oka. LC-MS determination of bioactive molecules
based upon stable isotope-coded derivatization method.
J. Pharm. Biomed. Anal. 2012, 69, 174.
[12] K. Guo, C. Ji, L. Li. Stable-isotope dimethylation labeling
combined with LCÀESI MS for quantification of amine-
containing metabolites in biological samples. Anal. Chem.
2007, 79, 8631.
A
novel isotope-labeling strategy was developed for
differential metabolomics analysis, using N-benzoyloxy-
succinimide. The derivatization reaction was selective
towards amine, phenol and thiol groups, with possibilities
for multiple derivatization on certain metabolites. In
comparison with their unlabeled counterparts, derivatized
metabolites exhibit significant increases in sensitivity and
retention time, allowing sufficient retention for polar
metabolites and satisfactory separation of isomers (e.g., Leu
and Ile). The reagent was prepared in two isotopic forms
(light and heavy), the heavy version containing six 13C
atoms. In contrast with deuterium-containing isotopic
analogs, the presence of 13C prevented isotope effects
and therefore ensured optimal relative quantitation. The
quantitative performance was investigated with an
experimental design using biological samples (cell extracts).
Method accuracy was investigated by comparing 115 singly
labeled feature pairs for peak area ratios (light/heavy
derivative) in cell extracts with known concentration ratios,
and an accuracy median bias of –5.7%. In addition, precision
(CV) of peak area ratios (n = 5) was <25% for over75 % of the
peak pairs.
The method was applied to a cellular model of oxidative
stress to assess its application for characterizing metabolite
changes in biological samples. Cell cultures (n = 5) treated with
1 mM H2O2 for 20 min and untreated cell cultures (n = 5) were,
respectively, derivatized with the light and heavy reagent.
Seventy-seven features were deemed discriminant (p <0.05,
change >50%). Structural elucidation was carried out based
on accurate mass determination, MS/MS spectra, and
comparison with standard model metabolites, resulting in ten
discriminating metabolites being clearly identified, with
fold-changes ranging from –2.6 to +4.0.
[13] C. Ji, W. Li, X. D. Ren, A. F. El-Kattan, R. Kozak, S. Fountain,
C. Lepsy. Diethylation labeling combined with UPLC/MS/
MS for simultaneous determination of
a panel of
monoamine neurotransmitters in rat prefrontal cortex
microdialysates. Anal. Chem. 2008, 80, 9195.
[14] K. Guo, L. Li. Differential C12/C13-isotope dansylation
labeling
and
fast
liquid
chromatography/mass
spectrometry for absolute and relative quantification of the
metabolome. Anal. Chem. 2009, 81, 3919.
Rapid Commun. Mass Spectrom. 2015, 29, 1632–1640
Copyright © 2015 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/rcm

M. Wagner et al.
[15] R. K. Zhou, K. Guo, L. Li. 5-Diethylaminonaphthalene-1-
sulfonyl chloride (DensCl): A novel triplex isotope labeling
reagent for quantitative metabolome analysis by liquid
chromatography mass spectrometry. Anal. Chem. 2013, 85,
11532.
[25] Z.-J. Zhu, A. W. Schultz, J. Wang, C. H. Johnson, S. M. Yannone,
G. J. Patti, G. Siuzdak. Liquid chromatography quadrupole
time-of-flight mass spectrometry characterization of
metabolites guided by the METLIN Database. Nat. Protocols
2013, 8, 451.
[26] D. S. Wishart, D. Tzur, C. Knox, R. Eisner, A. C. Guo,
N. Young, D. Cheng, K. Jewell, D. Arndt, S. Sawhney,
C. Fung, L. Nikolai, M. Lewis, M. A. Coutouly, I. Forsythe,
P. Tang, S. Shrivastava, K. Jeroncic, P. Stothard, G. Amegbey,
D. Block, D. D. Hau, J. Wagner, J. Miniaci, M. Clements,
M. Gebremedhin, N. Guo, Y. Zhang, G. E. Duggan,
G. D. Macinnis, A. M. Weljie, R. Dowlatabadi, F. Bamforth,
D. Clive, R. Greiner, L. Li, T. Marrie, B. D. Sykes, H. J. Vogel,
L. Querengesser. HMDB: the Human Metabolome
Database. Nucleic Acids Res. 2007, 35, D521.
[16] N. Abello, P. P. Geurink, M. van der Toorn, A. J. van Oosterhout,
J. Lugtenburg, G. A. van der Marel, H. A. Kerstjens,
D. S. Postma, H. S. Overkleeft, R. Bischoff. Poly(ethylene
glycol)-based stable isotope labeling reagents for the
quantitative analysis of low molecular weight metabolites
by LCÀMS. Anal. Chem. 2008, 80, 9171.
[17] M. R. Shortreed, S. M. Lamos, B. L. Frey, M. F. Phillips,
M. Patel, P. J. Belshaw, L. M. Smith. Ionizable isotopic labeling
reagent for relative quantification of amine metabolites by
mass spectrometry. Anal. Chem. 2006, 78, 6398.
[18] K. Shimbo, A. Yahashi, K. Hirayama, M. Nakazawa,
H. Miyano. Multifunctional and highly sensitive precolumn
reagents for amino acids in liquid chromatography/tandem
mass spectrometry. Anal. Chem. 2009, 81, 5172.
[19] D. W. Johnson. Free amino acid quantification by LC–
MS/MS using derivatization generated isotope-labelled
standards. J. Chromatogr. B 2011, 879, 1345.
[27] D. S. Wishart, T. Jewison, A. C. Guo, M. Wilson, C. Knox,
Y. Liu, Y. Djoumbou, R. Mandal, F. Aziat, E. Dong,
S. Bouatra, I. Sinelnikov, D. Arndt, J. Xia, P. Liu, F. Yallou,
T. Bjorndahl, R. Perez-Pineiro, R. Eisner, F. Allen, V. Neveu,
R. Greiner, A. Scalbert. HMDB 3.0 – The Human Metabolome
Database in 2013. Nucleic Acids Res. 2013, 41, D801.
[28] C. Montalbetti, V. Falque. Amide bond formation and
peptide coupling. Tetrahedron. 2005, 61, 10827.
[20] W.-C. Yang, H. Mirzaei, X. Liu, F. E. Regnier. Enhancement
of amino acid detection and quantification by electrospray
ionization mass spectrometry. Anal. Chem. 2006, 78, 4702.
[21] W. Yuan, J. Zhang, S. Li, J. L. Edwards. Amine metabolomics
of hyperglycemic endothelial cells using capillary LC–MS
with isobaric tagging. J. Proteome Res. 2011, 10, 5242.
[22] K. Guo, J. Peng, R. Zhou, L. Li. Ion-pairing reversed-phase
liquid chromatography fractionation in combination with
isotope labeling reversed-phase liquid chromatography-
mass spectrometry for comprehensive metabolome
profiling. J. Chromatogr. A 2011, 1218, 3689.
[23] J. M. Zheng, R. A. Dixon, L. Li. Development of isotope
labeling LC-MS for human salivary metabolomics and
application to profiling metabolome changes associated with
mild cognitive impairment. Anal. Chem. 2012, 84, 10802.
[24] C. A. Smith, G. O’Maille, E. J. Want, C. Qin, S. A. Trauger,
T. R. Brandon, D. E. Custodio, R. Abagyan, G. Siuzdak.
METLIN: a metabolite mass spectral database. Ther. Drug
Monit. 2005, 27, 747.
[29] M. Turowski, N. Yamakawa, J. Meller, K. Kimata,
T. Ikegami, K. Hosoya, N. Tanaka, E. R. Thornton.
Deuterium isotope effects on hydrophobic interactions:
The importance of dispersion interactions in the
hydrophobic phase. J. Am. Chem. Soc. 2003, 125, 13836.
[30] H. Shibasaki, T. Furuta, Y. Kasuya. Quantification of
corticosteroids in human plasma by liquid chromatography–
thermospray mass spectrometry using stable isotope dilution.
J. Chromatogr. B 1997, 692, 7.
[31] R. J. Zhang, C. S. Sioma, S. H. Wang, F. E. Regnier.
Fractionation of isotopically labeled peptides in quantitative
proteomics. Anal. Chem. 2001, 73, 5142.
SUPPORTING INFORMATION
Additional supporting information can be found in the online
version of this article at the publisher's website.
wileyonlinelibrary.com/journal/rcm
Copyright © 2015 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2015, 29, 1632–1640