Regioselective solvent-phase deuteration of polyphenolic compounds informs their identification by mass spectrometry

Article abstract of DOI:10.1016/j.ab.2014.02.018

Liquid chromatography-mass spectrometry (LC-MS) is a highly sensitive tool for the analysis of polyphenolic compounds in complex food and beverage matrices. However, the high degree of isomerism among polyphenols in general often complicates this approach, especially for identification of novel compounds. Here, we explore the utility of mild acid-catalyzed deuterium (MACD) labeling via electrophilic aromatic substitution as a complementary method for informing polyphenolic compound structure elucidation. To prevent hydrolysis of acid-labile glycosidic linkages, optimal reaction conditions that maximize regioselective hydrogen/deuterium (H/D) exchange of aromatic protons while preserving compound integrity were characterized (60°C, pH 3.0, 72 h). Under these conditions, standard compounds varying in the number and position of hydroxyl, glycosyl, and methyl groups about their aromatic core structure produced distinguishable H/D exchange patterns. The applicability of this method for the analysis of complex mixtures was demonstrated in red wine where the extent of deuterium exchange, together with accurate mass information, led to the putative identification of an unknown compound. The identification was further supported by tandem MS (MS/MS) data, which matched conclusively to the same compound in the Metlin LC-MS/MS library. With the capacity to discriminate between select isomeric forms, MACD labeling provides structural information that complements accurate mass and tandem mass spectral measurements for informing the identification of polyphenolics by MS.

Full text of DOI:10.1016/j.ab.2014.02.018

Analytical Biochemistry 452 (2014) 76–85
Contents lists available at ScienceDirect
Analytical Biochemistry
journal homepage: www.elsevier.com/locate/yabio
Regioselective solvent-phase deuteration of polyphenolic compounds
informs their identification by mass spectrometry
Mikel R. Roe a,b, Jerry D. Cohen a,b, Adrian D. Hegeman a,b,c,
⇑
a Department of Horticultural Science, University of Minnesota–Twin Cities, St. Paul, MN 55108, USA
b Microbial and Plant Genomics Institute, University of Minnesota–Twin Cities, St. Paul, MN 55108, USA
c Department of Plant Biology, University of Minnesota–Twin Cities, St. Paul, MN 55108, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 13 July 2013
Received in revised form 14 February 2014
Accepted 15 February 2014
Available online 26 February 2014
Liquid chromatography–mass spectrometry (LC–MS) is a highly sensitive tool for the analysis of polyphe-
nolic compounds in complex food and beverage matrices. However, the high degree of isomerism among
polyphenols in general often complicates this approach, especially for identification of novel compounds.
Here, we explore the utility of mild acid-catalyzed deuterium (MACD) labeling via electrophilic aromatic
substitution as a complementary method for informing polyphenolic compound structure elucidation. To
prevent hydrolysis of acid-labile glycosidic linkages, optimal reaction conditions that maximize regiose-
lective hydrogen/deuterium (H/D) exchange of aromatic protons while preserving compound integrity
were characterized (60 °C, pH 3.0, 72 h). Under these conditions, standard compounds varying in the
number and position of hydroxyl, glycosyl, and methyl groups about their aromatic core structure pro-
duced distinguishable H/D exchange patterns. The applicability of this method for the analysis of complex
mixtures was demonstrated in red wine where the extent of deuterium exchange, together with accurate
mass information, led to the putative identification of an unknown compound. The identification was fur-
ther supported by tandem MS (MS/MS) data, which matched conclusively to the same compound in the
Metlin LC–MS/MS library. With the capacity to discriminate between select isomeric forms, MACD label-
ing provides structural information that complements accurate mass and tandem mass spectral measure-
ments for informing the identification of polyphenolics by MS.
Keywords:
Metabolomics
Polyphenolics
Flavonoid
Stilbenoid
Deuterium labeling
Mass spectrometry
Compound identification
Ó 2014 Elsevier Inc. All rights reserved.
Polyphenolic compounds are secondary metabolites from the
phenylpropanoid biosynthetic pathway in plants that are often
present in plant-derived foods, beverages and dietary supplements
[1,2]. Polyphenolics often demonstrate phytoalexin and antioxi-
dant properties important for promoting abiotic and biotic stress
tolerance. In humans, these same properties are likely responsible
for the reported health benefits associated with consuming poly-
phenolics, namely decreased incidence of coronary heart disease,
various cancers, and neurodegeneration [3–10].
Of the various polyphenolic classes commonly found in food
and beverages (e.g., flavonoids, stilbenoids, chalcones), flavonoids
are the most abundant and chemically diverse. According to struc-
ture–activity relationship studies, the antioxidant, anti-inflamma-
tory, and substrate binding activities of flavonoids are dependent
on a number of key chemical features about their C6-C3-C6 diaro-
matic flavan core [11–14]. These features, including a hydroxyl
group at the C3 position, a ketone group at the C4 position, a dou-
ble bond between C2 and C3, and the attachment of the benzene
ring (B-ring)1 to either the C2 or C3 position, describe various states
of oxidation about the central heterocyclic pyran ring (C-ring) and
define the various subclasses of flavonoids, namely flavonols, flav-
ones, isoflavones, flavanones, flavan-3-ols, and anthocyanidins [11–
13]. Variable degrees and patterns of glycosylation, methylation,
and acylation further increase the structural complexity of dietary
flavonoids, which also greatly affects their bioavailability and thus
their therapeutic potential [1,15]. In all, more than 7000 unique fla-
vonoid structures have been characterized in sources such as fruits,
1
Abbreviations used: LC, liquid chromatography; MS, mass spectrometry; DAD,
diode array detector; UV, ultraviolet; NMR, nuclear magnetic resonance; H/D,
hydrogen/deuterium; D2O, deuterium oxide; CI, chemical ionization; MACD, mild
acid-catalyzed deuterium; CH3OD, deuterated methanol-d; DCOOD, deuterated
formic acid-d2; SPE, solid-phase extraction; MS/MS, tandem MS; CID, collision-
induced dissociation; RPLC, reverse-phase LC; UHPLC, ultra-high-performance liquid
chromatography; HCD, higher energy collision-induced dissociation; NCE, normalized
collision energy; ESI, electrospray ionization.
⇑
Corresponding author at: Department of Horticultural Science, University of
Minnesota–Twin Cities, St. Paul, MN 55108, USA.
E-mail address: hegem007@umn.edu (A.D. Hegeman).
http://dx.doi.org/10.1016/j.ab.2014.02.018
0003-2697/Ó 2014 Elsevier Inc. All rights reserved.

Regioselective deuteration of polyphenolics / M.R. Roe et al. / Anal. Biochem. 452 (2014) 76–85
77
vegetables, grains, and their derived beverages, and new compounds
continue to be identified with the characterization of novel
plant sources and the development of advanced analytical
techniques [16].
rium (MACD) labeling of polyphenolic compounds for informing
their identification by MS. Importantly, the conditions used for
MACD largely preserve compound integrity while producing
sufficient H/D exchange information to enhance structural
identifications.
Due to the complexity of biologically derived samples and the
low abundance of many polyphenolics, liquid chromatography–
mass spectrometry (LC–MS) is often the analytical platform of
choice for characterizing polyphenolics in complex biological mix-
tures. It is routinely conducted with orthogonal detection methods
that provide complementary information for discriminating be-
tween the large number of isomeric structures that comprise this
class of compounds [17,18]. The most typical approach involves
placing an ultraviolet diode array detector (DAD) upstream of the
mass analyzer (LC–UV–MS), thereby providing additional informa-
tion regarding the A-ring and C-ring aglycone structures [17,19–
22]. However, although UV absorbance sufficiently discriminates
among the various aglycone structures, it is less reliable for assign-
ing the position of methyl, glycoside, and acyl substituents about a
given aglycone [17]. Conversely, the coupling of LC–MS with nucle-
ar magnetic resonance (LC–MS–NMR) has shown during recent
years to be a powerful technique for unequivocally elucidating
polyphenolic structures within mixtures [23]. Although LC–MS–
NMR readily discriminates among many regio- and stereoisomers
unresolved by UV, it requires MS-compatible deuterated solvents,
which not only are costly but also may complicate interpretation
of the mass spectrum due to variable rates of deuterium exchange
of ‘‘active’’ protons throughout the structure. In addition, NMR
spectroscopy is far less sensitive than mass spectrometric analysis,
which often limits the complementarity of NMR data to only a
small subset of the ionized species.
Materials and methods
Chemicals
Apigenin, (+)-catechin, daidzein, fisetin, genistein, isoorientin,
morin, myricetin, orientin, naringenin, phloretin, piceid, quercetin,
quercitrin, rhamnetin, and resveratrol were purchased from Sig-
ma–Aldrich (St. Louis, MO, USA), and astragalin, epicatechin, epi-
catechin gallate, isoquercitrin, kaempferol, populnin, luteolin, and
cynaroside were purchased from Extrasynthese (Genay, France).
Solvents were obtained from Sigma–Aldrich and included CHRO-
MASOLV Plus HPLC (high-performance liquid chromatography)-
grade acetonitrile, formic acid (>97%), deuterated methanol-d
(CH3OD, 99 atom % D), and deuterated formic acid-d2 (DCOOD,
98 atom % D). D2O (99 atom % D) was purchased from Cambridge
Isotope Laboratories (Andover, MA, USA). BioUltra-grade ammo-
nium acetate was also obtained from Sigma–Aldrich. Monovarietal
red wine (2009) vinted from cold-climate hardy Frontenac grapes
was kindly provided by the University of Minnesota Horticultural
Research Center.
MACD labeling
One alternative, gas-phase hydrogen/deuterium (H/D) ex-
change, has been used to inform flavonoid structure by character-
izing the number and position of acidic flavonoid protons, both at
OH groups and at nucleophilic aromatic positions on the A- and B-
rings [24–26]. For instance, the Brodbelt group exposed flavonoid
ions to deuterium oxide (D2O) gas in an ion trap and demonstrated
how the resulting H/D exchange patterns, as well as exchange
kinetics, could be used to discriminate various flavonoids, includ-
ing stereoisomeric structures [24,25]. Although powerful, the
widespread use of their method has been limited because it re-
quires an in-house D2O gas delivery system into the trap portion
of the instrument. By comparison, H/D exchange of flavonoids dur-
ing ionization in a chemical ionization (CI) source has also been
shown to generate variable labeling patterns of various flavonoids,
with exchange occurring at all OH groups as well as aromatic pro-
tons in positions ortho and para to OH groups [26]. The utility of
this H/D exchange method is also limited because CI is classically
coupled to gas chromatography, a mode of sample fractionation
not typical for flavonoid analysis. Still, these studies well illustrate
the utility of H/D exchange as a complementary technique with MS
for profiling polyphenolics in complex mixtures and point to the
need for development of more approachable H/D exchange
methodologies.
Solution-phase H/D exchange is a potentially useful alternative
to gas-phase approaches for characterizing polyphenolics because
it does not require specialized equipment and is compatible with
liquid chromatographic fractionation of complex mixtures. How-
ever, we are not aware of such methods having been described
in the current literature. The numerous solution-phase H/D ex-
change protocols for generating deuterated polyphenolic standards
are typically designed for quantification studies and, thus, employ
relatively harsh pH and temperature conditions to maximize the
completeness of labeling [27–34]. Such conditions are inappropri-
ate for informing the structure of biological flavonoids because
they perturb compound integrity via hydrolysis of glycosidic and
acyl linkages. Here, we describe offline mild acid-catalyzed deute-
Standard compounds
In general, methanolic stocks (1 mg/ml) of each standard com-
pound were prepared in 2.0-ml microcentrifuge tubes. From these
stocks, aliquots were transferred to 1.5-ml microcentrifuge tubes
and concentrated to dryness via vacuum centrifugation. The result-
ing pellets were reconstituted in deuterated methanol (10% of final
volume) and diluted to volume with MACD labeling buffer (D2O
containing 10 mM ammonium formate, pH 3.0). The samples were
then incubated in the dark for 72 h at 60 °C and then frozen at
ꢀ20 °C to stop the labeling reaction. Labeling controls were also
generated by reacting each compound in nondeuterated buffer
(H2O containing 10 mM ammonium formate, pH 3.0). In addition
to these standard conditions, various MACD labeling buffers as well
as reaction times and temperatures were also evaluated in the time
course of atom % incorporation and compound degradation studies
described below (see Fig. 3). Except for the analysis of quercetin
and catechin in Figs. 1 and 2 (see figures below in Results and Dis-
cussion), respectively, all reactions were conducted in triplicate
and analyzed independently.
Red wine
Freshly opened Frontenac wine (40 ml) was poured into a 50-ml
conical tube and vigorously vortexed to homogenize the sample. As
a labeling control, apigenin, an exogenous flavonoid foreign to red
wines, was added to a final concentration of 10 lg/ml. From this
stock, two 1-ml aliquots were transferred to 2-ml microcentrifuge
tubes and concentrated to dryness via vacuum centrifugation. To
generate MACD-labeled wine, one of the resulting pellets was
reconstituted in 1 ml of MACD labeling buffer (D2O containing
10 mM ammonium acetate, pH 3.0) to regenerate the original 1ꢁ
wine concentration. The other pellet, serving as an unlabeled con-
trol, was reconstituted in nondeuterated buffer (H2O containing
10 mM ammonium acetate, pH 3.0). The two samples were then
incubated in the dark for 72 h at 60 °C and subsequently stored
at ꢀ20 °C to minimize further labeling.

78
Regioselective deuteration of polyphenolics / M.R. Roe et al. / Anal. Biochem. 452 (2014) 76–85
Quercetin: deuterated versus unlabeled
Deuterated quercetin (2 ml, 20 g/ml), prepared according to
the MACD labeling protocol described above, was buffer exchanged
via C18 solid-phase extraction (SPE; SepPak, Waters) and eluted
into acetonitrile containing 20% D2O. Specifically, the SepPak car-
tridge was pretreated with acetonitrile (20% [v/v] D2O), equili-
tions were performed in triplicate, and 25-
lected at 2, 4, 8, 24, 48, 72, and 96 h after initiation. A 5-
ll aliquots were col-
ll
l
aliquot from each time point was subsequently analyzed by re-
verse-phase LC–MS on an LTQ-Orbitrap XL mass spectrometer
(ThermoFisher) using the solvents and flow rate described in the
‘‘LC–MS’’ section below. Solvents were combined to generate the
following mobile-phase gradient: 0 to 1 min, 20% B; 1 to 4 min,
20 to 30% B; 4 to 6 min, 35 to 45% B; 6 to 8 min, 90% B; 8 to
10 min, 20% B. Additional instrument parameters are described in
the ‘‘LC–MS’’ section below. Atom % incorporation was determined
empirically by normalizing the isotope pattern of labeled com-
pounds to that of their unlabeled form. Degradation was monitored
for each compound by measuring the change in integrated peak
area as a function of time. Details of the approach used may be
found below in the ‘‘Data analysis’’ section.
brated with 100% D2O, loaded with 40 lg of deuterated quercetin
in 2 ml of MACD labeling buffer, and washed with 100% D2O prior
to elution with 20% (v/v) D2O in acetonitrile. The eluted quercetin
was subsequently evaporated to dryness under vacuum and recon-
stituted in 1 ml of cold (4 °C) 10 mM ammonium formate buffer
(pH 5.0 in H2O) to promote the selective H/D back exchange from
quercetin’s OH groups. Both MS and tandem MS (MS/MS) spectra
were collected for deuterated and unlabeled quercetin on a triple
quadrupole mass spectrometer (TSQ Quantum Access, Thermo-
Fisher) in the negative ionization mode. The samples, 40
l
g/ml
Structure-dependent H/D exchange patterns of MACD-labeled
polyphenolic standards
each, were infused at a flow rate of 20 l/min and detected using
l
the following instrument parameters: spray voltage, 3.5 kV; vapor-
izer temperature, 50 °C; sheath gas, 20 psi; auxiliary gas, 0 psi. Col-
lision-induced dissociation (CID) fragmentation spectra were also
collected for each compound using a collision energy setting of
25 eV.
Three stock mixtures (A, B, and C) containing different phenolic
compounds were prepared to 100 lg/ml in methanol for each
compound. Stock A contained quercetin, phloretin, luteolin-8-C-
glucoside (orientin), epicatechin, kaempferol, resveratrol, dihyr-
oxycinnamic acid, and myricetin. Stock B contained luteolin,
naringenin, daidzein, fisetin, luteolin-6-glucoside (isoorientin),
kaempferol-7-glucoside, quercetin-3-glucoside, and resveratrol-5-
glucoside (piceid). Stock C contained rhamnetin, genistein, epicat-
echin gallate, morin, apigenin, and quercetin-3-rhamnoside.
H/D back exchange of catechin during reverse-phase LC–MS and C18
SPE?
To evaluate the H/D back exchange of OH groups as well as the
A-ring of flavonoids during LC–MS, the flavan-3-ol catechin, a fla-
vonoid with a particularly reactive A-ring for H/D exchange, was
MACD labeled as described above. Specifically, 1 ml of catechin
Aliquots (80
tuted in 5
with D2O to regenerate 100
was subsequently subjected to MACD labeling conditions in tripli-
cate by combining 20 l of each stock with 20 l of 100 mM
ammonium formate (pH 3.0 in D2O) and 160 l of D2O. The reac-
tion mixtures were incubated at 60 °C in the dark; aliquots
(20 l) were removed over time (2, 4, 8, 24, 48, and 72 h) and
stored at ꢀ20 °C prior to analysis. Samples (5 l) from each time
l
l) of each stock were dried under vacuum, reconsti-
l of dimethyl sulfoxide (DMSO), and diluted to 80
g/ml for each compound. Each stock
l
ll
l
(50 lg/ml) in MACD labeling buffer was incubated for 24 h at
60 °C. Half of the resulting deuterium-labeled catechin was buffer
exchanged into acetonitrile with 20% D2O via C18 SPE as described
above for quercetin. The resulting eluate was concentrated to dry-
l
l
l
ness via vacuum centrifugation and reconstituted in 500
ll of cold
l
H2O with 10% methanol. Both the non-buffer-exchanged and buf-
fer-exchanged MACD-labeled catechin, together with non-H/D-ex-
l
point and from unexchanged solutions of each stock were subse-
quently analyzed by LC–MS using an LTQ-Orbitrap XL mass spec-
trometer and the flow rate and solvents described below in the
‘‘LC–MS’’ section. Solvents were combined to generate the follow-
ing mobile-phase gradient: 0 to 3 min, 20 to 30% B; 3 to 4 min,
30 to 40% B; 4 to 6 min, 40% B; 6 to 7 min, 40 to 45% B; 7 to
8 min, 45 to 90% B; 8 to 10 min, 90% B; 10 to 13 min, 20% B. The
number of H/D-exchanged positions, as well as the atom % incor-
poration, was determined empirically by normalizing the isotope
pattern of labeled compounds to that of their unlabeled form.
The details of this analysis can be found below in the ‘‘Data analy-
sis’’ section.
changed catechin (50
into a triple quadrupole mass spectrometer as described above
for quercetin. In addition, a 5- l aliquot of the non-buffer-ex-
lg/ml in H2O), were subsequently infused
l
changed MACD-labeled catechin was also analyzed by LC–MS on
the same instrument using the following parameters: spray volt-
age, 3.5 kV; vaporizer temperature, 50 °C; sheath gas, 60 psi; aux-
iliary gas, 5 psi; mass range, 285 to 300 m/z. To promote H/D back
exchange of deuterated catechin on column, 90% H2O with 10%
acetonitrile was passed through the column at 400 ll/min for
20 min prior to compound elution in 90% acetonitrile with 10%
H2O. Averaged mass spectra comprising 20 scans were generated
for each sample and visually compared.
LC–MS
Time course of atom % incorporation and compound degradation
Standard compounds
A
standard mixture composed of quercetin-3-rhamnoside
All LC–MS analyses of standard compounds were conducted by
reverse-phase LC (RPLC) on an HSS T3 C18 UPLC (ultra-performance
(quercitrin), kaempferol-3-glucoside (astragalin), luteolin-7-gluco-
side (cynaroside), resveratrol-5-glucoside (piceid), and epicatechin
gallate was dried via vacuum centrifugation and reconstituted in
liquid chromatography) column (2.1 mm i.d. ꢁ 100 mm, 1.8
lm
particle size; Waters) using a flow rate of 400 l/min. Organic sol-
l
70
1330
To compare the extent of deuterium incorporation and compound
degradation under various acidic conditions, three 40- l aliquots
of the stock solution were diluted 10-fold into D2O solvent and buf-
fered to pH 3.0 with 10 mM ammonium formate and pH 4.0 and
5.0 with 10 mM ammonium acetate. Exchange reactions at pHs
3.0, 4.0, and 5.0 were conducted at 60 and 90 °C, with an additional
set of pH 3.0 reactions also conducted at 30 °C. All exchange reac-
l
l of CH3OD. The dissolved sample was then diluted with
vent gradients were generated with an Accela ultra-high-perfor-
mance liquid chromatography (UHPLC) pump (ThermoFisher) by
combining 10 mM ammonium acetate (pH 5.0 in H2O; solvent A)
and 100% acetonitrile (solvent B) according to the programs de-
scribed above for each experiment. Unit resolution measurements
on quercetin labeling and catechin back exchange were conducted
on a triple quadrupole mass spectrometer (TSQ Quantum Access,
ThermoFisher) using the instrument parameters described above
for the respective experiments. The accurate mass measurements
l
l of D2O to generate a 100- g/ml compound stock solution.
l
l

Regioselective deuteration of polyphenolics / M.R. Roe et al. / Anal. Biochem. 452 (2014) 76–85
79
conducted for all remaining experiments were made on an LTQ-
Orbitrap XL using the following parameters: negative mode ioniza-
tion, 7500 resolution, 500 ls trap fill time, 1 microscan, 4.0 kV
spray voltage, 60 psi nitrogen sheath gas, 10 psi nitrogen auxiliary
gas, 280 °C capillary temperature, and 90 V tube lens voltage. Data
were collected in profile mode across a range of 150 to 650 m/z.
compositions. For this approach, the ratio of the monoisotopic
ion to the +1 and +2 ions, respectively, was first measured for each
compound in its natural abundance (unlabeled) form. These ratios
were then used as a template and applied to the isotopic channels
of the deuterated compounds, beginning with the monoisotopic
ion, to measure the fractional contribution made by each subpop-
ulation of differentially labeled compounds (0D, 1D, 2D, etc.) to the
whole. For instance, to measure the relative amount of unlabeled
compound remaining postlabeling, the intensity of the monoiso-
topic ion and the corresponding amounts of signal from the +1
and +2 channels were totaled and divided by the total signal across
the isotopic envelope. These intensities were then subtracted from
the isotopic envelope to generate an adjusted spectrum. This pro-
cess was repeated iteratively across the isotopic envelope until
the maximum number of labeled sites was found. Using the frac-
tional contribution made by each subpopulation (0D, 1D, 2D,
etc.) to the whole, the atom % deuterium incorporation for each la-
beled compound was calculated by multiplying the ratio of actual-
to-maximal number of H/D-exchanged positions by 100.
Frontenac red wine
Frontenac red wine was analyzed on a hybrid quadrupole-Orbi-
trap (Q Exactive) mass spectrometer (ThermoFisher) equipped
with an Ultimate 3000 UHPLC pump (ThermoFisher). The wine
was chromatographically resolved at a flow rate of 300
ll/min on
a
reverse-phase C18 column (HSS T3, 2.1 mm i.d. ꢁ 100 mm,
1.8 lm particle size; Waters) by mixing mobile-phase solvent A
(10 mM ammonium acetate, pH 5.0) and solvent B (100% acetoni-
trile) to generate the following gradient: 0 to 2 min, 5% B; 2 to
22 min, 5 to 35% B; 22 to 24 min, 35% B; 24 to 27 min, 35 to 90%
B; 27 to 32 min, 90% B; 32 to 32.5 min, 90 to 5% B; 32.5 to
40 min, 5% B. Accurate mass measurements (70,000 resolution)
from 200 to 2000 m/z were collected in negative ionization mode
using the following parameters: 4.0 kV spray voltage, 60 psi
nitrogen sheath gas, 10 psi nitrogen auxiliary gas, 50 °C vaporizer
temperature, 280 °C capillary temperature. Data-dependent tan-
dem mass spectra of the top 10 most abundant ions (5-s dynamic
exclusion) were generated in the higher energy collision-induced
dissociation (HCD) cell using a normalized collision energy (NCE)
setting of 40%.
Peak area integration and computing compound degradation
Generation and integration of extracted ion chromatograms
(EICs) was performed in batch using the Quan method within the
Processing Setup tool in Xcalibur (version 4.1, ThermoFisher). Spe-
cifically, five peaks corresponding to the modified forms of five
polyphenolics and a peak for the external standard catechin were
defined in a single Quan method using similar parameters as de-
scribed above. In addition, peaks were smoothed using a moving
average of 7 scans prior to integration. The resulting values were
reported using the Quan Peak Table output in the Non-Repeating
section found in the XReport tool within Xcalibur.
Data analysis
Average isotope patterns
Deuterium isotope effects are well known to produce significant
shifts in chromatographic retention [35]. To account for differences
in isotope peak distributions over a typical peak elution window,
isotope patterns were averaged across a retention window, includ-
ing 9 MS scans about the peak apex for each compound; these were
automatically computed and exported in batch using the Process-
ing Setup tool in Xcalibur (version 4.1, ThermoFisher). Individual
processing files for each compound were generated using the Quan
method with the following generic parameters tailored to each
compound: mass range, 6 m/z beginning with the quasi-molecular
or M + 1 ion for the nondeuterated compound; retention time,
60 s on either side of the previously observed elution time; sig-
nal-to-noise threshold, 3.0. In addition, report templates were gen-
erated for each compound’s Quan method using the XReport tool
within Xcalibur. Spectrum List Tables accessed through the Quan
Repeating section were created using the Quan method processing
parameters for each individual compound. The resulting .txt report
files for each compound were concatenated into a single .txt file
and further processed in Microsoft Excel (2007).
To measure compound degradation, the average integrated
peak area (n = 3) for each of the five modified polyphenolics at each
time point was normalized against that of the external standard
catechin. The ratio of the normalized signal at each time point rel-
ative to that at time point 0 h was then subtracted from 1 to yield
the fraction of hydrolyzed compound: 1 – (signal x h/signal 0 h).
Multiplying the resulting number by 100 provided the % hydroly-
sis, which was plotted against time to generate the time-depen-
dent hydrolysis profile for each compound.
Results and discussion
Polyphenolic compounds are a chemically diverse and abun-
dant class of plant-derived antioxidants that are commonly charac-
terized in complex biological matrices via MS. To enhance the
power of MS for identifying polyphenolics on a metabolomic scale,
we optimized a stable isotope labeling method, based on MACD
labeling, that introduces deuterium into polyphenolic compounds
in a structure-dependent manner. Polyphenolics incubated in
mildly acidified D2O at relatively low temperatures undergo H/D
exchange at aromatic protons ortho (C6 and C8) to resorcinolic hy-
droxyl groups (5 and 7 OH) via electrophilic aromatic substitution
(Scheme 1) [29]. The hydroxyl groups (OD) coincidently labeled
during this process are subsequently back exchanged (OH)
Atom % incorporation
To calculate atom % incorporation, an empirical approach was
adopted in preference to probabilistic calculations that rely on de-
fined elemental compositions so that the approach could be used
with unknown compounds or those with ambiguous elemental
Scheme 1. Mild acid-catalyzed deuterium (MACD) labeling of polyphenolic compound.

80
Regioselective deuteration of polyphenolics / M.R. Roe et al. / Anal. Biochem. 452 (2014) 76–85
following buffer exchange into H2O while the much less acidic aro-
matic protons remain stably deuterated.
-
H
OH
OH
A
B
C
0D
H
HO
H
O
To demonstrate MACD labeling of polyphenolics, the flavonol
quercetin was incubated in acidified D2O (pH 3.0) for 72 h at
60 °C and then exchanged into H2O via C18 SPE. The overlaid MS
spectra of unlabeled and MACD-labeled quercetin indicate hetero-
geneous labeling of the latter, with an estimated 55% of com-
pounds containing one D incorporation (302 m/z) and another
42% containing two incorporations (303 m/z) (Fig. 1A). In addition,
when the 303 m/z ion was subjected to CID MS/MS, only the A-ring
containing fragment ions were shifted by 2 m/z units relative to the
fragmentation spectrum of unlabeled quercetin (Fig. 1B). For
example, while the 0,4Aꢀ fragment shifted from 107 to 109 m/z,
the 1,2Bꢀ ion remained unchanged at 121 m/z. Together, these data
indicate that the conditions used for MACD labeling selectively ex-
change the A-ring aromatic protons at positions C6 and C8 and are
not harsh enough to activate H/D exchange at the aromatic protons
on the catecholic B-ring.
In light of the selective back exchange of quercetin OH groups
following C18 SPE, we sought to streamline sample processing by
removing the SPE step from the protocol and relying solely on RPLC
in protic solvent to induce the desired back exchange. For this pur-
pose, mass spectrometric analysis of MACD-labeled catechin (pH
3.0, 60 °C, 24 h), a flavonoid especially sensitive toward H/D ex-
change (or back exchange) at its C6 and C8 carbons, was conducted
in D2O labeling buffer or in protic solvent following either C18 SPE
or inline RPLC (Fig. 2). Compared with unlabeled catechin (289 m/
z; Fig. 2A), the isotope pattern for MACD-labeled catechin infused
in D2O reaction buffer (Fig. 2B) indicated a heterogeneous popula-
tion of labeled compounds, with the maximally labeled com-
pounds (295 m/z) corresponding to H/D exchange at all four
nonionized OH groups as well as the aromatic positions C6 and
C8 on the A-ring. However, when MACD-labeled catechin in D2O
reaction buffer was analyzed by LC–MS in protic solvents, the
OH
OH
-D
OD
OD
D
5D
DO
D
O
4D
6D
OD2D
2D
3D
OD
-
-
H
H
OH
OH
OH
D
HO
D
O
O
OH
1D
OH
OH
D
D
2D
HO
D
OH
1D
OH
289
291
293
295
297
m/z
Fig. 2. (ꢀ) ESI–MS spectra of unlabeled catechin infused in H2O (A) and stable
isotope-labeled catechin following MACD labeling and infused in D2O reaction
buffer (B), infused in H2O following C18 solid-phase extraction (SepPak) for buffer
exchange (C), or following reverse-phase liquid chromatography with the use of
H2O aqueous phase (D).
resulting compounds are largely homogeneously labeled with
two deuterium atoms (291 m/z), suggesting that the OH groups
back exchanged while positions C6 and C8 remained deuterated
(Fig. 2C). Notably, this isotope pattern was not affected by exposing
the labeled compound on-column to variable volumes of protic
solvent prior to detection (data not shown), further illustrating
the stability of the incorporation at these two positions on the A-
ring. A similar labeling pattern was observed for C18 SPE-processed
MACD-labeled catechin, indicating that RPLC was comparable to
SPE for the selective back exchange of polyphenolic OH groups
prior to their detection by MS (Fig. 2D).
With the goal of using MACD to inform the native polyphenolic
constituents in biological samples, we evaluated the impact that
various reaction conditions had on both the extent of deuterium
labeling and the stability of acid-labile glycosidic linkages for a pa-
nel of structurally distinct compounds (Fig. 3). In general, temper-
ature had a much greater impact than pH on deuterium labeling for
all compounds, with higher atom % incorporations for all time
points at 90 °C as compared with 60 and 30 °C (Fig. 3, solid line).
Conversely, pH-dependent labeling effects were observed only for
the stilbenoid resveratrol-5-glucoside (piceid), with pH 3.0 produc-
ing a modest increase in atom % incorporation over pHs 4.0 and 5.0
at both 60 and 90 °C (Fig. 3D). In addition, whereas the type of
sugar conjugated to the flavonoid aglycone appeared to be inconse-
quential for labeling (Fig. 3A vs. Fig. 3B), the position of the sugar
conjugate was not, with the atom % incorporation being greatly
reduced at both 60 and 90 °C for the 7-O-glucoside of luteolin com-
pared with the 3-O-glucoside of kaempferol (Fig. 3B vs. Fig. 3C).
Similar results were observed for compound integrity, with the
extent of glycoside hydrolysis, measured by the ratio of glycoside
to the external standard catechin, being noticeably greater at
90 °C than at 60 or 30 °C for all compounds (Fig. 3, dashed lines).
In addition, a measurable pH effect was also observed, most nota-
bly for quercetin-3-rhamnoside, where the most acidic condition
(pH 3.0) produced the greatest levels of hydrolysis. Compound
structure was also a strong determinant of glycoside hydrolysis,
with the type and position of the conjugated sugar greatly affecting
A
Unlabeled
Deuterated
100
50
100
50
0
0
299 301 303 305 307 m/z
299 301 303 305 307m/z
H/D
-H
HO
O
C
B
A
151/153
D/H
100
50
O
OH
H/D
-H
-H
-H
O
HO
H/D
A
OH
B
HO
-H
O
O
OH
B
HO
OH
A
H/D
A
D/H
O
D/H
O
OH
301/303
OH
D/H
OH
179/181
OH
107/109
121/121
273/275
0
m/z
Fig. 1. Mass spectral comparison of unlabeled (black) and stable isotope labeled
(blue) quercetin generated via MACD exchange. (A) Side-by-side comparison of the
isotope patterns appearing in the (ꢀ) ESI–MS spectra. (B) Overlaid MS/MS spectra
from the unlabeled and H/D-exchanged compound showing putative structures for
select CID fragments according to Fabre and coworkers [38]. (For interpretation of
the references to color in this figure legend, the reader is referred to the Web
version of this article.)

Regioselective deuteration of polyphenolics / M.R. Roe et al. / Anal. Biochem. 452 (2014) 76–85
81
A
Quercetin-3-rhamnoside
OH
100
75
50
25
0
H
O
% hydrolysis
X
HO
O
X
O
O
OH O
OH
HO
hr.
OH
0
2 4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96
30°C, pH 3
60°C, pH 3 60°C, pH 4 60°C, pH 5 90°C, pH 3 90°C, pH 4 90°C, pH 5
B
100
75
50
25
0
Kaempferol-3-glucoside
OH
X
% hydrolysis
HO
X
O
OH
OH
O
O
OH O
HO
OH
hr.
0
2 4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96
30°C, pH 3
60°C, pH 3 60°C, pH 4 60°C, pH 5 90°C, pH 3 90°C, pH 4 90°C, pH 5
C
Luteolin-7-glucoside
OH
100
75
50
25
0
OH
% hydrolysis
OH
HO
OH
X
O
O
X
O
OH
OH O
hr.
0
0
0
2 4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96
30°C, pH 3
60°C, pH 3 60°C, pH 4 60°C, pH 5 90°C, pH 3 90°C, pH 4 90°C, pH 5
D
Resveratrol-5-glucoside
100
75
50
25
0
OH
X
OH
% hydrolysis
O
X
HO
HO
O
X
OH
OH
hr.
2
4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96
30°C, pH 3
60°C, pH 3 60°C, pH 4 60°C, pH 5 90°C, pH 3 90°C, pH 4 90°C, pH 5
E
100
75
50
25
0
Epicatechin gallate
OH
% Hydrolysis
OH
X
HO
X
O
X
O
OH
OH
OH
O
hr.
2
4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96 0 2 4 8 24 48 72 96
X
OH
30°C, pH 3
60°C, pH 3 60°C, pH 4 60°C, pH 5 90°C, pH 3 90°C, pH 4 90°C, pH 5
Fig. 3. H/D exchange and glycoside hydrolysis for representative glycosylated polyphenolic compounds under various MACD labeling conditions. Time course plots of atom %
incorporation (solid line) and % hydrolysis (dashed line) for quercetin-3-rhamnoside (A), kaempferol-3-glucoside (B), luteolin-7-glucoside (C), resveratrol-5-glucoside (D),
and epicatechin gallate (E) are shown on the same y axis for seven different temperature/pH combinations. The positions susceptible to stable deuterium incorporation under
MACD labeling conditions are designated on each compound’s structure with a gray ‘‘X’’.
the integrity of the glycosidic linkage. For example, greater
amounts of hydrolysis were observed under most conditions for
the flavonol conjugated to rhamnose at the C3 position (querce-
tin-3-rhamnoside) than the one conjugated to glucose at the same
position (kaempferol-3-glucoside) (Fig. 3A vs. Fig. 3B). In addition,
comparison of kaempferol-3-glucoside with luteolin-7-glucoside
shows the disparate impact that glucoside position has on com-
pound integrity, with the former exhibiting much higher levels of
hydrolysis under all conditions (Fig. 3B vs. Fig. 3C). Interestingly,
glucose conjugated to the stilbene resveratrol was more resistant
to hydrolysis than when the sugar was conjugated to the C-ring
but not the A-ring of flavonoids (Fig. 3D vs. Fig. 3B and C).
under all conditions tested, with nearly complete labeling (50 atom
% incorporation) of the A-ring achieved by 8 h irrespective of tem-
perature or pH (Fig. 3E, solid line). Surprisingly, at 90 °C two pro-
tons on the gallate moiety also exchanged (atom % incorporation
>50%), showing a slightly faster exchange rate at higher pH. By
comparison, only two positional incorporations were observed
for the epicatechin aglycone alone at all temperatures, thereby
confirming the placement of the additional two exchanged posi-
tions for epicatechin gallate on the gallate moiety (data not
shown).
Considering that ester linkages, such as the one joining epicat-
echin and gallic acid, are known to be susceptible to acid-catalyzed
hydrolysis, we also investigated the degradation of epicatechin gal-
late during incubation in the various MACD labeling conditions de-
scribed above. Similar to the polyphenolic glycosides, epicatechin
gallate was degraded largely in a temperature-dependent manner,
In addition to glycosylated phenolics, we also monitored deute-
rium labeling and compound integrity for the acid-esterified
flavan-3-ol: epicatechin gallate (Fig. 3E). Compared with the previ-
ous compounds, epicatechin gallate was very sensitive to labeling

82
Regioselective deuteration of polyphenolics / M.R. Roe et al. / Anal. Biochem. 452 (2014) 76–85
with the highest levels occurring at 90 °C, followed by more mod-
est rates at 60 and 30 °C (Fig. 3E, dashed line). Additional pH-
dependent degradation at each temperature was negligible. Hence,
for all compounds tested, incubation at 90 °C produced both the
most robust deuterium labeling of aromatic protons and the most
compound degradation, whereas reactions conducted at 60 °C
resulted in more modest labeling but also reduced levels of com-
pound degradation.
Based on these findings, we further sought to evaluate how the
aglycone structure itself affects H/D exchange patterns under reac-
tion conditions that largely avoid degradation (i.e., 60 °C, pH 3.0,
72 h). A comparison of representative compounds from four differ-
ent phenolic classes revealed structure-dependent labeling pat-
terns in both the number of exchanged positions (outer bars) and
the completeness of labeling (inner bars), as represented by total
atom % incorporation (Fig. 4). Following a clear order of reactivity
(stilbenoid > chalcone > flavonoid ꢂ dihydroxycinnamic acid), the
number of exchanged positions generally corresponded with the
number of aromatic protons positioned ortho and para relative to
the resorcinolic OH groups (indicated by ‘‘X’’), whereas the degree
of deuterium labeling followed expected trends based on the num-
ber and position of ring-activating substituents. Chalcone (phlore-
tin) was the lone exception, undergoing H/D exchange at three
positions despite having only two such aromatic protons, a
following incubation at 60 °C (pH 3.0) for 72 h. In general, com-
pounds with unmodified resorcinol-like OH groups (meta-posi-
tioned to each other) were deuterated at more positions and
with greater atom % incorporation than their modified counter-
parts (Fig. 5). This discrepancy in labeling pattern was observed
in compounds lacking a 5-OH group in the A-ring (Fig. 5B, 1 vs.
2, 3 vs. 4), in compounds glycosylated and methylated at the 7-
OH of the A-ring (Fig. 5C, 3 vs. 5, 6 vs. 7), and in compounds such
as morin and myricetin where the presence of meta-positioned OH
groups in the B-ring corresponded to two additional sites of H/D
exchange compared with flavonols with phenolic (kaempferol)
and catecholic (quercetin) B-ring configurations (Fig. S3 [see online
supplementary material], 13 and 14 vs. 6 and 3). These findings
were corroborated in a pair of stilbenoids where glycosylation at
the 5-OH position resulted in a greatly reduced atom % incorpora-
tion compared with the unmodified form (Fig. 5C, 11 vs. 12). In
addition to OH groups, glycosylation of aromatic carbons ortho to
the A-ring OH groups in flavonoids (C6 and C8) also affected deu-
terium labeling patterns, with C6 and C8 glucosylated luteolin
being labeled at one fewer position than the unmodified luteolin
compound (Fig. 5D, 8 vs. 9 and 10).
The electronic configuration of the C-ring also affected H/D ex-
change patterns. For instance, a comparison of three compounds
(kaempferol, apigenin, and naringenin), which differ only in their
C-ring compositions, revealed that reduction of the double bond
between C2 and C3, as seen in the flavanone naringenin, not only
phenomenon likely due to additional labeling of the
a-carbons
via enolization [36]. Interestingly, deuterating the compounds for
shorter periods at higher temperatures (90 °C, pH 3.0) produced
time-dependent effects on the number of positions labeled as well
as the extent of labeling. For example, whereas the deuterium
labeling pattern after 8 h was similar to that observed for the 72-
h time point at 60 °C, additional sites of H/D exchange were gener-
ated for the flavonoid and chalcone following 24 h of incubation
(Fig. 4). However, caution must be taken when incubating com-
pounds at elevated temperatures because additional oxidation
and other undesirable side reactions will likely occur.
Flavonoid
Stilbenoid
A
R4
H
R3
R1
A
A
H
H
R2
R1
OH
11, 12
Given the abundance, high degree of structural variability, and
potent antioxidant properties of flavonoids, we sought to further
characterize the structure-dependent labeling patterns of this
important class of polyphenolic compounds under the conditions
described above. Specifically, the deuterium labeling patterns of
compounds varying in the number and position of hydroxyl, glyco-
syl and methyl groups about the three rings were compared
Compound
Genistein
R1 R2 R3 R4
1
OH
H
H
H
H
H
H
H
H
H
OH
OH
H
H
H
H
H
H
H
H
H
2
3
4
5
6
7
8
9
Daidzein
Quercetin
OH
H
OH
OH
OH
OH
OH
Fisetin
OH
Rhamnetin
MeO
OH
GlcO
OH
Kaempferol
Kaempferol-7-O-glucoside
Luteolin
Luteolin-6-C-glucoside
OH GlcC OH
10 Luteolin-8-C-glucoside
11 Resveratrol
OH
H
OH GlcC
OH
Flavonoid
OH
Stilbenoid
X
4
3
2
1
0
12 Resveratrol-5-glucoside
GlcO
OH
OH
X
B
C
D
HO
HO
X
O
O
X
X
3
2
1
0
OH
OH
HO
Dihydroxycinnamic acid
O
Dihydrochalcone
OH
X
X
OH
X
O
HO
X
OH
HO
OH
X
HO
60°C
(72 h)
1 2 3 4
3 5 6 7 1112
8 9 10
Compound (60°C / 72 h)
Fig. 4. Comparison of H/D exchange patterns for various phenolic compound
classes following MACD labeling at pH 3.0 under different temperature/time
reaction conditions. The internal colored bars indicate atom % incorporation for the
observed number of exchanged positions indicated by the outer noncolored bars.
Fig. 5. H/D exchange patterns for select flavonoids and stilbenoids with A-ring
compositions (A) that vary based on the presence of an OH group at R1 (B), O-
conjugation of the OH group at R3 (flavonoids) or R1 (stilbenoids) (C), and C-
conjugation of R2 or R4 (D). MACD labeling was conducted at 60 °C (pH 3.0) for
72 h. The internal colored bars indicate % atom incorporation for the observed
number of exchanged positions indicated by the outer noncolored bars. Error bars
for atom % incorporation reflect standard deviations from three replicates. MeO, O-
methylation; GlcO, O-glucoside; GlcC, C-glucoside.
Error bars for atom
replicates. The structures of the compounds examined are presented in the legend,
with gray ‘‘X’’ marking the predicted sites of H/D exchange based on the
% incorporation reflect standard deviations from three
a
orientation of substituents about the aromatic rings. Flavonoid: quercetin; dihydr-
ochalcone: phloretin; stilbenoid: resveratrol; dihydroxycinnamic acid: 2,4-dihyr-
oxycinnamic acid.

Regioselective deuteration of polyphenolics / M.R. Roe et al. / Anal. Biochem. 452 (2014) 76–85
83
disrupts the conjugated pi system through the ring but also yields
one additional deuterium incorporation into the compound
(Fig. 6A). Similar to the chalcone described earlier, this additional
D-exchanged positions and the atom % incorporation rendered
the labeling patterns less distinguishable from those of their
unmodified counterparts (Fig. S2). This loss of sensitivity to struc-
tural variation is most clearly seen in the 24-h reaction data. To-
gether, these data indicate that the H/D-exchanged patterns of
flavonoids generated via select MACD labeling conditions are sen-
sitive to structural variations and can be used to complement other
mass spectral data to inform compound structure.
site of H/D exchange likely occurred at the
a-carbon adjacent to
the ketone via acid-catalyzed enolization [36]. Further supporting
this assertion is the fact that epicatechin, a compound lacking both
a double bond between C2 and C3 and a ketone at C4, produced the
same number of exchanged positions as kaempferol and apigenin,
compounds with conjugated C-rings (Fig. 6, 19 vs. 6 and 15). In
addition to the nature of the C2/C3 bond, the type of substituent
attached to the C3 OH group also affected the deuterium labeling
pattern of polyphenolic compounds. For example, whereas glyco-
sylation of quercetin’s 3-OH group did not change the number of
exchanged positions (Fig. 6B), esterification of this same group in
epicatechin by gallic acid resulted in two additional sites of H/D ex-
change (Fig. 6C). These additional sites likely occur at each of the
ortho carbons on the phenolic ring of the gallate substituent and,
according to the atom % incorporation for this compound (54%),
are not robustly labeled under the MACD labeling conditions used.
As seen with the representative phenolic compounds discussed
previously (Fig. 4), changing the reaction conditions to higher
temperature (90 °C, pH 3.0) and shorter incubation periods (8
and 24 h) also altered the observed H/D-exchanged patterns of
flavonoids in a compound-specific manner (see Figs. S1–S4 of
supplementary material). The greatest change occurred in A-ring
modified compounds, where increases in both the number of H/
To aid in the interpretation of H/D exchange patterns resulting
from MACD labeling, we established a set of general principles
based on the regioselective deuterium labeling patterns of
polyphenolic compounds described. When conducting MACD
labeling at 60 °C (pH 3.0) for 72 h, H/D exchange occurs at (i) each
aromatic position adjacent to meta-oriented hydroxyl groups, such
as C6 and C8 in the A-ring of most flavonoids and C2 and C6 of gal-
lic acid conjugates, and at (ii) sp3 hybridized carbons adjacent (
a)
to carbonyls, such as C3 in the C-ring of flavanones. However, the
number of exchangeable positions is decreased by one if either
OH group in the meta-oriented pair is modified via methylation
or glycosylation or if any of the adjacent aromatic positions is c-
glycosylated or dimerized. Paired with accurate mass information,
these principles may greatly reduce the complexity in discriminat-
ing between isomeric polyphenolic compounds that share similar
structures.
To illustrate the potential of MACD labeling for discriminating
between isomeric flavonoids in complex mixtures, we applied
A
I.S.
A
4
320.05
100
269/271
100
O
C
O
C
321.05
3
2
1
0
319.04
R1
18
301/303
O
O
318.04
322.06
317.03
0
Compound
Kaempferol
R1
OH
H
m/z
6
50
15 Apigenin
317/321
18 Naringenin
N/A
6 15 18
3 16 17
19 20
3
2
O
B
0
C
4
6
8
10 12 14 16 18 20 22 24
R1
Time (min)
O
B
1
0
Compound
Quercetin
R1
norwedelic acid
gossypetin
3
OH
16 Quercetin-3-O-glucoside
17 Quercetin-3-O-rhamnoside
GlcO
RhamO
C15H10O 8
quercetagetin
5'-hydroxymorin
myricetin
C
Gal = gallic acid
5
4
3
2
O
O
hexahydroxyflavone
bitalgenin
O
C
R1
HO
OH
0
1
2
3
4
5
OH
Predicted number of H/D
exchangeable positions
1
0
Compound
R1
19 Epicatechin
20 Epicatechin gallate
OH
GalO
Fig. 7. MACD-labeled Frontenac red wine. (A) Overlaid base peak chromatograms of
unlabeled (black) and MACD-labeled (blue) Frontenac red wine with representative
labeled ion pairs, including those for the internal standard (I.S.) apigenin, marked
with the nominal mass for both the unlabeled and MACD-labeled ions. The isotope
pattern for the labeled form of the 317 m/z ion is also shown (inset). (B) Predicted
number of H/D-exchanged positions for isomeric flavonoids (C15H10O8) in the Arita
laboratory flavonoid database that match by mass to the 317.0303 m/z (1 ppm mass
error) ion [16]. Hexahydroxyflavone: 3,5,8,30,40,50-hexahydroxyflavone. (For inter-
pretation of the references to color in this figure legend, the reader is referred to the
Web version of this article.)
Fig. 6. H/D exchange patterns generated for select flavonoids with C-ring compo-
sitions that vary based on the presence of an OH group at R1 or a double bond
between C2 and C3 (A), O-conjugation of the OH group at R1 (B), and O-
esterification by gallic acid at R1 (C). MACD labeling was conducted at 60 °C (pH
3.0) for 72 h. The internal colored bars indicate % atom incorporation for the
observed number of exchanged positions indicated by the outer noncolored bars.
Error bars for atom
% incorporation reflect standard deviations from three
replicates. GlcO, O-glucoside; RhamO, O-rhamnoside; GalO, O-gallate.

84
Regioselective deuteration of polyphenolics / M.R. Roe et al. / Anal. Biochem. 452 (2014) 76–85
our method to the analysis of red wine. As shown in Fig. 7, the
overlaid chromatograms of the unlabeled (black) and deuterated
(blue) samples are identical, indicating that the complexity of the
wine sample was not differentially perturbed by the varying sol-
vent treatments. More important, many of the ions were shifted
in mass following MACD labeling, including the internal standard
apigenin (269 m/z) as well as unknowns at 301 and 317 m/z. That
the mass shift varied among the labeled ions (2 vs. 4 Da) suggests
that MACD labeling in the wine sample was sensitive to structural
differences between compounds. Moreover, the extent of deute-
rium labeling observed for apigenin is consistent with the value
predicted based on its structure, demonstrating that the MACD
labeling rules defined using standard compounds apply to the
analysis of complex mixtures as well.
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.ab.2014.02.018.
References
[1] C. Manach, G. Williamson, C. Morand, A. Scalbert, C. Rémésy, Bioavailability
and bioefficacy of polyphenols in humans: I. Review of 97 bioavailability
studies, Am. J. Clin. Nutr. 81 (2005) 230S–242S.
[2] S. Quideau, D. Deffieux, C. Douat-Casassus, L. Pouységu, Plant polyphenols:
chemical properties, biological activities, and synthesis, Angew. Chem. Int. Ed.
50 (2011) 586–621.
[3] P. Barberger-Gateau, C. Raffaitin, L. Letenneur, C. Berr, C. Tzourio, J.F. Dartigues,
A. Alpérovitch, Dietary patterns and risk of dementia: the three-city cohort
study, Neurology 69 (2007) 1921–1930.
[4] Q. Dai, A.R. Borenstein, Y. Wu, J.C. Jackson, E.B. Larson, Fruit and vegetable
juices and Alzheimer’s disease: the Kame project, Am. J. Med. 119 (2006) 751–
759.
[5] P. Fresco, F. Borges, C. Diniz, M.P. Marques, New insights on the anticancer
properties of dietary polyphenols, Med. Res. Rev. 26 (2006) 747–766.
[6] A. Gomes, E. Fernandes, J.L. Lima, L. Mira, M.L. Corvo, Molecular mechanisms of
anti-inflammatory activity mediated by flavonoids, Curr. Med. Chem. 15
(2008) 1586–1605.
[7] J.D. Lambert, J. Hong, G. Yang, J. Liao, C.S. Yang, Inhibition of carcinogenesis by
polyphenols: evidence from laboratory investigations, Am. J. Clin. Nutr. 81
(2005) 284S–291S.
[8] M.R. Peluso, Flavonoids attenuate cardiovascular disease, inhibit
phosphodiesterase, and modulate lipid homeostasis in adipose tissue and
liver, Exp. Biol. Med. 231 (2006) 1287–1299.
[9] C.S. Yang, J.M. Landau, M.-T. Huang, H.L. Newmark, Inhibition of carcinogenesis
by dietary polyphenolic compounds, Annu. Rev. Nutr. 21 (2001) 381–406.
[10] G. Williamson, C. Manach, Bioavailability and bioefficacy of polyphenols in
humans: II. Review of 93 intervention studies, Am. J. Clin. Nutr. 81 (2005)
243S–255S.
[11] D. Amic´, D. Davidovic´-Amic´, D. Beslo, V. Rastija, B. Lucic´, N. Trinajstic´, SAR and
QSAR of the antioxidant activity of flavonoids, Curr. Med. Chem. 14 (2007)
827–845.
[12] J.S. Choi, H.Y. Chung, S.S. Kang, M.J. Jung, J.W. Kim, J.K. No, H.A. Jung, The
structure–activity relationship of flavonoids as scavengers of peroxynitrite,
Phytother. Res. 16 (2002) 232–235.
[13] P. Cos, L. Ying, M. Calomme, J.P. Hu, K. Cimanga, B. Van Poel, L. Pieters, A.J.
Vlietinck, D. Vanden, Berghe, structureꢀactivity relationship and classification
of flavonoids as inhibitors of xanthine oxidase and superoxide scavengers, J.
Nat. Prod. 61 (1998) 71–76.
[14] J.J. van Zanden, H.M. Wortelboer, S. Bijlsma, A. Punt, M. Usta, P.J. van Bladeren,
I.M. Rietjens, N.H. Cnubben, Quantitative structure–activity relationship
studies on the flavonoid mediated inhibition of multidrug resistance
proteins 1 and 2, Biochem. Pharmacol. 69 (2005) 699–708.
[15] C.S. Yang, S. Sang, J.D. Lambert, M.-J. Lee, Bioavailability issues in studying the
health effects of plant polyphenolic compounds, Mol. Nutr. Food Res. 52
(2008) S139–S151.
[16] M. Arita, K. Suwa, Search extension transforms Wiki into a relational system: a
case for flavonoid metabolite database, BioData Min. 1 (2008) 7.
[17] E. de Rijke, P. Out, W.M.A. Niessen, F. Ariese, C. Gooijer, U.A.T. Brinkman,
Analytical separation and detection methods for flavonoids, J. Chromatogr. A
1112 (2006) 31–63.
Although the base peak for many labeled ions represented the
maximally labeled species, in some instances it did not, as seen
in the spectrum for the labeled form of the 317 m/z ion (Fig. 7A, in-
set). In this case, the prominent +4 isotopologue (321 m/z), mea-
suring roughly 70% of the base peak ion, clearly indicated the
presence of compounds with four exchanged positions. The value
of this information for informing the identity of the 317 m/z ion
is demonstrated in Fig. 7B. Of the seven isomeric compounds in
the Arita flavonoid database that match based on mass alone
(1 ppm mass error), only one (myricetin) is predicted to incorpo-
rate four deuterium atoms according to its structure (black bar). In-
deed, the fragmentation spectrum of the 317 m/z ion matched well
with the reference spectrum for myricetin in the Metlin LC–MS/MS
database, further legitimizing the utility of the MACD labeling data
to inform compound structure [37] (Fig. S5).
MACD labeling of polyphenolic compounds is a largely nonde-
structive chemical labeling strategy that uses electrophilic aro-
matic substitution to drive H/D exchange at positions ortho and
para to meta-oriented phenolic OH groups such as the A-ring of
most flavonoids and the B-ring of certain flavonoids. Compared
with existing protocols that employ harsher reaction conditions,
MACD labeling is much more sensitive to changes in compound
structure that alter the electrophilic properties of polyphenolics,
including the number, orientation, and modification status of hy-
droxyl groups. It is especially useful for discriminating between
positional isomers of glycosylated flavonoids that are variably
modified at positions involved in electrophilic aromatic substitu-
tion such as meta-oriented OH groups and their ortho carbons.
Importantly, the sensitivity of the MACD labeling method is suffi-
cient for profiling polyphenolics in complex biological matrices,
as demonstrated here in the analysis of red wine, and is expected
to largely depend on sample preparation and the analytical limita-
tions of the MS platform employed for analysis. In conclusion,
when combined with accurate mass and a comprehensive flavo-
noid database, the extent of H/D exchange generated from MACD
labeling can be used to inform the identification of unknown poly-
phenolic compounds in complex mixtures. In addition, it should
have utility for characterizing the polyphenolic profiles of novel
biological samples where it may aid in the structural elucidation
of previously uncharacterized compounds.
[18] P.F. Pinheiro, J.C. Goncalo, Structural analysis of flavonoids and related
compounds:
Phytochem:
a
A
review of spectroscoptic applications, in: V. Rao (Ed.),
Global Perspective of Their Role in Nutrition and Health,
InTech, 2012, pp. 33–56.
[19] G. Aldini, L. Regazzoni, A. Pedretti, M. Carini, S.-M. Cho, K.-M. Park, K.J. Yeum,
An integrated high resolution mass spectrometric and informatics approach
for the rapid identification of phenolics in plant extract, J. Chromatogr. A 1218
(2011) 2856–2864.
[20] I. Nicoletti, C. Bello, A. De Rossi, D. Corradini, Identification and quantification
of phenolic compounds in grapes by HPLC–PDA–ESI–MS on a semimicro
separation scale, J. Agric. Food Chem. 56 (2008) 8801–8808.
[21] L.-Z. Lin, J.M. Harnly, A screening method for the identification of glycosylated
flavonoids and other phenolic compounds using
a standard analytical
Acknowledgments
approach for all plant materials, J. Agric. Food Chem. 55 (2007) 1084–1096.
[22] Z. Liang, C.L. Owens, G.-Y. Zhong, L. Cheng, Polyphenolic profiles detected in
the ripe berries of Vitis vinifera germplasm, Food Chem. 129 (2011) 940–950.
[23] M.V. Silva Elipe, Advantages and disadvantages of nuclear magnetic resonance
spectroscopy as a hyphenated technique, Anal. Chim. Acta 497 (2003) 1–25.
[24] J. Zhang, J.S. Brodbelt, Gas-phase hydrogen/deuterium exchange and
conformations of deprotonated flavonoids and gas-phase acidities of
flavonoids, J. Am. Chem. Soc. 126 (2004) 5906–5919.
[25] E.D. Niemeyer, J.S. Brodbelt, Isomeric differentiation of green tea catechins
using gas-phase hydrogen/deuterium exchange reactions, J. Am. Soc. Mass
Spectrom. 18 (2007) 1749–1759.
[26] K.P. Madhusudanan, K. Sachdev, D.A. Harrison, D.K. Kulshreshtha, Gas phase
hydrogen/deuterium exchange reactions in chemical ionization mass spectra
of flavonoids, Int. J. Mass Spectrom. Ion Process. 62 (1984) 289–295.
This work was funded by National Science Foundation (NSF)
Plant Genome Research Program grants IOS-0923960 and IOS-
1238812. We thank the University of Minnesota Office of the Vice
President for Research and the Grant in Aid program for salary sup-
port (M.R.R.). We further acknowledge supplemental support from
the Gordon and Margaret Bailey Endowment for Environmental
Horticulture and the Minnesota Agricultural Experiment Station.
We also thank Gary Gardner (University of Minnesota) for helpful
discussions throughout the project.

Regioselective deuteration of polyphenolics / M.R. Roe et al. / Anal. Biochem. 452 (2014) 76–85
85
[27] S. Rasku, K. Wähälä, Synthesis of deuterium labeled polyhydroxy flavones and
3-flavonols, Tetrahedron 56 (2000) 913–916.
[28] S. Rasku, K. Wähälä, J. Koskimies, T. Hase, Synthesis of isoflavonoid
deuterium labeled polyphenolic phytoestrogens, Tetrahedron 55 (1999)
3445–3454.
[33] M. Jordheim, T. Fossen, J. Songstad, Ø.M. Andersen, Reactivity of anthocyanins
and pyranoanthocyanins: studies on aromatic hydrogen–deuterium exchange
reactions in methanol, J. Agric. Food Chem. 55 (2007) 8261–8268.
[34] K. Hiraoka, T. Miyamoto, S. Baba, T. Furuta, Preparation of deuterium-labeled
rutin by hydrogen exchange reaction, J. Labelled Compd. Radiopharm. 18
(1981) 613–619.
[29] K. Wähälä, S. Rasku, K. Parikka, Deuterated phytoestrogen flavonoids
and isoflavonoids for quantitation, J. Chromatogr.
122.
[30] U. Hakala, K. Wähälä, Expedient deuterolabeling of polyphenols in ionic
liquids–DCl/D2O under microwave irradiation, J. Org. Chem. 72 (2007) 5817–
5819.
B
777 (2002) 111–
[35] A. Valleix, S. Carrat, C. Caussignac, E. Léonce, A. Tchapla, Secondary isotope
effects in liquid chromatography behaviour of 2H and 3H labelled solutes and
solvents, J. Chromatogr. A 1116 (2006) 109–126.
[36] G.E. Lienhard, T.-C. Wang, On the mechanism of acid-catalyzed enolization of
ketones, J. Am. Chem. Soc. 91 (1969) 1146–1153.
[31] S. Faizi, H. Siddiqi, A. Naz, S. Bano, Lubna, Specific deuteration in patuletin and
related flavonoids via keto–enol tautomerism: solvent- and temperature-
dependent 1H-NMR studies, Helv. Chim. Acta 93 (2010) 466–481.
[32] T. Stark, N. Wollmann, S. Lösch, T. Hofmann, Quantitation of resveratrol in red
wines by means of stable isotope dilution analysisꢀultra-performance liquid
chromatographyꢀQuan-time-of-flight mass spectrometry and cross
validation, Anal. Chem. 83 (2011) 3398–3405.
[37] C. Smith, G. O’Maille, E. Want, C. Qin, S. Trauger, T. Brandon, D.E. Custodio, R.
Abagyan, G. Siuzdak, METLIN:
a metabolite mass spectral database—
therapeutic drug monitoring, Ther. Drug Monit. 27 (2005) 747–751.
[38] N. Fabre, I. Rustan, E. de Hoffmann, J. Quetin-Leclercq, Determination of
flavone, flavonol, and flavanone aglycones by negative ion liquid
chromatography electrospray ion trap mass spectrometry, J. Am. Soc. Mass
Spectrom. 12 (2001) 707–715.