Development of a QuEChERS-Based Stable-Isotope Dilution LC-MS/MS Method to Quantitate Ferulic Acid and Its Main Microbial and Hepatic Metabolites in Milk

Article abstract of DOI:10.1021/acs.jafc.6b03318

Forage plants of the Poaceae family are grown as pasturage or used for the production of hay, straw, corn stover, etc. Although ferulic acid contents of grasses are generally high, the amount of ingested ferulic acid differs depending on the type of forage, resulting in varying contents of ferulic acid and its microbial and hepatic metabolites in milk. Concentrations and patterns of these metabolites may be used as markers to track different forages in livestock feeding. Therefore, we developed a stable isotope dilution assay to quantitate ferulic acid, 12 ferulic acid-based metabolites, p-coumaric acid, and cinnamic acid in milk. Because most analytes were not commercially available as stable isotope labeled standard compounds, they were synthesized as ¹³C- or deuterium-labeled standard compounds. A modification of the QuEChERS method, a Quick, Easy, Cheap, Effective, Rugged, and Safe approach usually applied to analyze pesticides in plant-based products, was used to extract the phenolic acids from milk. Determination was carried out by LC-ESI-MS/MS in scheduled multiple reaction monitoring modus. By using three different milk samples, the applicability of the validated approach was demonstrated.

Full text of DOI:10.1021/acs.jafc.6b03318

Article
pubs.acs.org/JAFC
Development of a QuEChERS-Based Stable-Isotope Dilution LC-MS/
MS Method To Quantitate Ferulic Acid and Its Main Microbial and
Hepatic Metabolites in Milk
Martin Waterstraat, Andreas Hildebrand, Margit Rosler, and Mirko Bunzel*
Department of Food Chemistry and Phytochemistry, Karlsruhe Institute of Technology (KIT), Adenauerring 20a, 76131 Karlsruhe,
Germany
ABSTRACT: Forage plants of the Poaceae family are grown as pasturage or used for the production of hay, straw, corn stover,
etc. Although ferulic acid contents of grasses are generally high, the amount of ingested ferulic acid differs depending on the type
of forage, resulting in varying contents of ferulic acid and its microbial and hepatic metabolites in milk. Concentrations and
patterns of these metabolites may be used as markers to track different forages in livestock feeding. Therefore, we developed a
stable isotope dilution assay to quantitate ferulic acid, 12 ferulic acid-based metabolites, p-coumaric acid, and cinnamic acid in
milk. Because most analytes were not commercially available as stable isotope labeled standard compounds, they were
synthesized as 13C- or deuterium-labeled standard compounds. A modification of the QuEChERS method, a Quick, Easy, Cheap,
Effective, Rugged, and Safe approach usually applied to analyze pesticides in plant-based products, was used to extract the
phenolic acids from milk. Determination was carried out by LC-ESI-MS/MS in scheduled multiple reaction monitoring modus.
By using three different milk samples, the applicability of the validated approach was demonstrated.
KEYWORDS: phenolic acids, hydroxycinnamic acids, ferulic acid, metabolism, milk, QuEChERS, isotopic labeling,
stable isotope dilution assay, mass spectrometry
example, p-coumaric (p-CA) and cinnamic acid (CinnA, Figure
1), as well as from polyphenols such as flavonoids and from the
aromatic amino acids phenylalanine and tyrosine. These
phenolic compounds also naturally occur in plant materials,
and their microbial and/or hepatic metabolism partially results in
the same metabolites as FA.22,23
INTRODUCTION
■
Ferulic acid (4-hydroxy-3-methoxycinammic acid, FA) largely
contributes to the stability of plant cell walls of grasses and other
plants. As a potential cell wall cross-link it is also a main factor in
cell wall recalcitrance against chemical or enzymatic degradation.
Larger amounts of FA can be found in grasses (members of the
family Poaceae), especially in maize (Zea mays), whereas
quantities in dicotyledonous plants of the order of Caryophyl-
lales are generally lower.1−3 Only small amounts of FA occur in
its free form, but most FA is linked to cell wall polymers, i.e.
arabinoxylans, pectins, and/or lignin.4−6 These polymers can be
cross-linked by ferulic acid and its derivatives, such as ferulate
oligomers, limiting cell growth and increasing the robustness of
the cellulose−hemicellulose network.7,8
With many ruminant forages being based on grasses, dairy
cows ingest comparably large amounts of FA. Microbial esterases
in the rumen can liberate FA; other enzymes are able to reduce
the unsaturated aliphatic side-chain of FA and to demethylate
and dehydroxylate its phenolic unit (Figure 1, solid arrows).9−11
Microbial decarboxylation or acetate elimination can transform
FA to 4-vinylguaiacol or vanillin, respectively.12,13 Also, phenyl-
acetic acid (PAA), which is potentially formed from phenyl-
propionic acid (PPA) by microbial α-oxidation, was found in
milk.13−15 FA and its microbial metabolites are partially absorbed
and further metabolized, for example through hepatic β-
oxidation (Figure 1, dotted arrows).16,17 Through phase-II
metabolism, benzoic acids are conjugated with glycine to form
hippuric acids.13,18 Additionally, conjugates with sulfuric or
glucuronic acids are potential phase-II metabolites.18,19 Most of
these compounds were detected in cow’s milk earlier.20,21 It has
to be emphasized that the analyzed metabolites are not
exclusively formed from FA, but may also originate from, for
The amounts of FA in different forages vary considerably.
Consequently, the concentrations of FA and its metabolites in
milk potentially differ largely, depending on the cow’s feeding
type. For example, feeding maize silage should result in a higher
FA intake if compared to many other grasses or products thereof.
Also, feeding pure roughages compared to roughages with the
addition of concentrates might result in differences of the
phenolic acid profiles of milk. Thus, specific phenolic acids may
be used as markers to trace the forage types of dairy cows.
Increased consumer expectations on the potential health effects
of organic milk or milk with particular labels as well as the
demand of ethical husbandry have become important social
issues. However, currently, cattle forages are not routinely traced
back. In the past, fatty acid profiles were proposed as potential
indicators for different forage sources.24
Here, we report the development of a stable isotope dilution
assay to quantitate phenolic acids in milk as potential markers for
different cattle forages. QuEChERS-based sample extraction25,26
in combination with LC-MS/MS analysis provides an accurate,
precise, selective, and highly sensitive method. Whereas the
QuEChERS-based sample preparation simplifies and speeds up
Received: July 25, 2016
Revised: October 13, 2016
Accepted: October 15, 2016
© XXXX American Chemical Society
A
DOI: 10.1021/acs.jafc.6b03318
J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry
Article
Figure 1. Potential microbial (solid arrows) and hepatic (dotted arrows) metabolic pathways of ferulic acid (FA), p-coumaric acid (p-CA), and cinnamic
acid (CinnA) in ruminants. BA, benzoic acid; CaffA, caffeic acid; diHFA, dihydroferulic acid; 3,4-diOHPPA, 3-(3,4-dihydroxyphenyl)propionic acid;
HA, hippuric acid; 3-OHBA, 3-hydroxybenzoic acid; 4-OHBA, 4-hydroxybenzoic acid; 3-OHHA, 3-hydroxyhippuric acid; 3-OHPPA, 3-(3-
hydroxyphenyl)propionic acid; 4-OHPPA, 3-(4-hydroxyphenyl)propionic acid; PAA, phenylacetic acid; PPA, 3-phenylpropionic acid.
Figure 2. Scheme for the syntheses of 13C-labeled standard compounds. Benzaldehydes were converted to cinnamic acids by using the Knoevenagel
reaction (I; pyridine, aniline). Hydrogenation of the propenylic side-chains (II; methanol, palladium (10%) on carbon, H2) resulted in 3-
phenylpropionic acids. Asterisks indicate 13C-labels. 1, vanillin; 2, 3,4-dihydroxybenzaldehyde; 3, 3-hydroxybenzaldehyde; 4, 4-hydroxybenzaldehyde; 5,
benzaldehyde; 6, 3-hydroxycinnamic acid; CaffA, caffeic acid; CinnA, cinnamic acid; diHFA, dihydroferulic acid; 3,4-diOHPPA, 3-(3,4-
dihydroxyphenyl)propionic acid; FA, ferulic acid; 3-OHPPA, 3-(3-hydroxyphenyl)propionic acid; 4-OHPPA, 3-(4-hydroxyphenyl)propionic acid;
PPA, 3-phenylpropionic acid; p-CA, p-coumaric acid.
benzaldehyde (99%), dihydroferulic acid (diHFA, 97%), 3-(3,4-
dihydroxyphenyl)propionic acid (3,4-diOHPPA, 98%), 3-hydroxyben-
zoic acid (3-OHBA, 99%), 4-hydroxybenzoic acid (4-OHBA, 99%), 3-
(3-hydroxyphenyl)propionic acid (3-OHPPA, 98%), 3-(4-
hydroxyphenyl)propionic acid (4-OHPPA, 99%), and phenylpropionic
acid (PPA, 99%) were from Alfa Aesar (Haverhill, Massachusetts, USA).
Benzoic acid (BA, 99.5%), N,O-bis(trimethylsilyl)trifluoroacetamide
(BSTFA, 99%), caffeic acid (CaffA, 97%), CinnA (99%), 3-
hydroxybenzaldehyde (95%), glycine (99%), p-CA (98%), and vanillin
(99%) were from Fluka (Buchs, Switzerland), and benzoyl chloride
(99%), hippuric acid (HA, 99%), acetonitrile, methanol, magnesium
sulfate, pyridine, and toluol were from VWR International (Radnor,
Pennsylvania, USA). 4-Hydroxybenzaldehyde (96%), formic acid, and
the procedure, using isotopically labeled standard compounds
improves both the accuracy and precision of this method by
compensating for, for example, analyte losses during sample
cleanup procedures and matrix induced suppression of
ionization.
MATERIALS AND METHODS
■
Chemicals. [13C2]Glycine (99%), [13C3]malonic acid (99%),
acetone-d6 (99.9%), deuterium (99.8%), deuterium chloride 35 wt %
(99%), 3,4-dihydroxybenzaldehyde (97%), ethyl acetate, trans-FA
(99%), PAA (99%), and sodium hydroxide 40 wt % (99.5%) were
purchased from Sigma-Aldrich (St. Louis, Missouri, USA). Aniline,
B
DOI: 10.1021/acs.jafc.6b03318
J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry
Article
palladium (10%) on activated carbon were from Merck KGaA
(Darmstadt, Germany), [3,5-D2]BA (98.9%) was from CDN isotopes
[8,9-13C2]trans-p-Coumaric Acid. starting material, 60 mg (0.56
mmol) of [13C3]malonic acid, 112.4 mg (0.92 mmol) of 4-
hydroxybenzaldehyde, 300 μL of pyridine, 30 μL of aniline; product,
72.6 mg (0.44 mmol, yield 78%) of [13C2]p-CA. δH [ppm] 6.33 (ddd, J =
160.6; 15.9; 2.7 Hz, 1H, 8); 6.90 (m, 2H, 3;5); 7.55 (m, 2H, 2;6); 7.61
(ddd, J = 15.9; 6.8; 2.9 Hz, 1H, 7).
(Queb
́
ec, Canada), D2O was from Deutero GmbH (Kastellaun,
Germany), thionyl chloride (98%) was from Riedel-de Haen AG
(Seelze, Germany), and hydrogen was produced in-house by using a
hydrogen generator.
̈
Milk Samples. Conventional milk samples with 1.5% and 3.5% fat
contents (UHT, homogenized) were purchased in April 2016 from a
local grocery store (Rheinland-Pfalz, Germany). An organic milk sample
with a fat content of 3.8% (pasteurized, nonhomogenized) was also
[8,9-13C2]trans-3-Hydroxycinnamic Acid. starting material, 30
mg (0.28 mmol) of [13C3]malonic acid, 56.2 mg (0.46 mmol) of
[13C2]3-hydroxybenzaldehyde, 150 μL of pyridine, 15 μL of aniline;
product, 40.2 mg (0.24 mmol, yield 86%) of [13C2]3-hydroxycinnamic
acid.
purchased in April 2016 from a local dairy farm (Baden-Wurttemberg,
̈
Germany), which produces exclusively organic milk-products following
demeter-principles.
[8,9-13C2]trans-Cinnamic Acid. starting material, 60 mg (0.28
mmol) of [13C3]malonic acid, 97.6 mg (0.92 mmol) of benzaldehyde,
300 μL of pyridine, 30 μL of aniline; product, 44.8 mg (0.3 mmol, yield
54%) of [13C2]CinnA. δH [ppm] 6.53 (ddd, J = 161.3; 16.0; 2.7 Hz, 1H,
8); 7.44 (m, 3H, 3;4;5); 7.68 (m, 3H, 2;6;7).
1
Nuclear Magnetic Resonance Spectroscopy. H NMR experi-
ments were performed at 298 K on a Bruker (Rheinstetten, Germany)
Ascend 500 MHz NMR spectrometer equipped with a Prodigy
cryoprobe using standard Bruker implementations. Samples were
dissolved in acetone-d6 (carboxylic and phenolic protons were partially
exchanged for deuterium), and the acetone residual peak was used for
spectrum calibration (methyl proton, δH 2.05 ppm).27
Preparative Liquid Chromatography. Preparative separations
were performed on a Shimadzu (Kyoto, Japan) HPLC equipped with
two pumps (LC-8A), an UV-detector (SPD-20A), and a communica-
tion bus module (CBM-20A). Injection was accomplished manually via
a six valve port with a 2 mL sample loop.
LC-DAD-MS/MS. The LC-DAD-MS/MS consisted of a 2690
separations module, pumps, degasser, autosampler, and a 996 diode
array detector (Waters Corporation (Milford, Massachusetts, USA))
coupled to a Micromass Quattro Micro mass spectrometer (Waters
Corporation) and a column oven (Jetstream Plus, Beckman Coulter,
Krefeld, Germany).
Synthesis of Phenylpropionic Acids. Syntheses of 13C-labeled 3-
phenylpropionic acids [13C2]diHFA, [13C2]3,4-diOHPPA, [13C2]4-
OHPPA, [13C2]3-OHPPA, and [13C2]PPA were carried out by
hydrogenation of the corresponding carbon-labeled cinnamic acids
(Figure 2). The cinnamic acids (amounts see below) were reduced in 1.5
mL of methanol using palladium (10%) on carbon (approximately 5
mg) as catalyst under H2 atmosphere. After 16 h, the catalyst was
removed by centrifugation, and the solvent was evaporated. Structure
and purity of the products were evaluated by LC-DAD-MS and NMR as
described above.
[8,9-13C2]Dihydroferulic Acid. 49.1 mg (0.25 mmol) of [13C2]FA
was reduced to 47.3 mg (0.24 mmol, yield 96%) of [13C2]diHFA. δH
[ppm] 2.54 (ddt, J = 128.2; 7.3; 7.3 Hz, 2H, 8); 3.80 (s, 3H, OCH3);
6.66 (dd, J = 8.0; 1.9 Hz, 1H, 6); 6.72 (d, J = 8.0 Hz, 1H, 5); 6.84 (d, J =
1.9 Hz, 1H, 2).
Synthesis of Cinnamic Acids. Syntheses of the 13C-labeled
cinnamic acids [13C2]FA, [13C2]CaffA, [13C2]p-CA, [13C2]3-hydrox-
ycinnamic acid, and [13C2]CinnA were performed by using the
Knoevenagel reaction with Doebner modification (Figure 2), similar
to a formerly published approach.28 In brief, the respective
benzaldehyde was added to [13C3]malonic acid, pyridine, and aniline
(amounts see below) in a 2 mL glass vial. Prevented from light, the
sealed vial was heated to 55 °C for 16 h. Cold water (1 mL) was slowly
added, and the solution was acidified by adding dropwise 0.5 mL of
concentrated hydrochloric acid. The reaction product was transferred to
a larger vial and extracted into ethyl acetate (5 × 2 mL). The combined
extracts were evaporated to dryness using a rotary evaporator.
Purification of the reaction products by preparative HPLC using a
phenyl-hexyl column (Phenomenex, Aschaffenburg, Germany, Luna,
250 × 10 mm, 5 μm) yielded the standard compounds in sufficient
purity (>98%): Methanolic solutions of the raw products were
fractionated at room temperature with a flow rate of 3.5 mL min−1
and a detection wavelength of 280 nm; water with 0.1% formic acid (A)
and methanol with 0.1% formic acid (B) were used as eluents. Starting
conditions for the separation of [13C2]FA, [13C2]CaffA, [13C2]p-CA, and
[13C2]3-hydroxycinnamic acid used 20% B; the eluent was ramped to
45% B in 25 min and ramped to 100% B within 10 min, following an
equilibration step. To purify [13C2]CinnA, 35% B was initially used,
followed by a linear increase to 100% B within 35 min and an
equilibration step. The purified, dried compounds were analyzed with
LC-DAD-MS to determine their purity. In addition, 5 mg of the
compounds were dissolved in acetone-d6 to acquire 1H NMR spectra.
[8,9-13C2]trans-Ferulic Acid. starting material, 65 mg (0.61 mmol)
of [13C3]malonic acid, 142.8 mg (0.92 mmol) of vanillin, 300 μL of
pyridine, 30 μL of aniline; product, 85.2 mg (0.43 mmol, yield 71%) of
[13C2]FA. δH [ppm] 3.92 (s, 3H, OCH3); 6.37 (ddd, J = 160.8; 15.9; 2.7
Hz, 1H, 8); 6.87 (d, J = 8.2 Hz, 1H, 5); 7.14 (dd, J = 8.2; 2.0 Hz, 1H, 6);
7.33 (d, J = 2.0 Hz, 1H, 2); 7.59 (ddd, J = 15.9; 6.8; 2.9 Hz, 1H, 7).
[8,9-13C2]trans-Caffeic Acid. starting material, 65 mg (0.61 mmol)
of [13C3]malonic acid, 127 mg (0.92 mmol) of 3,4-dihydroxybenzalde-
hyde, 300 μL of pyridine, 30 μL of aniline; product, 79.1 mg (0.43 mmol,
yield 71%) of [13C2]CaffA. δH [ppm] 6.26 (ddd, J = 160.4; 15.9; 2.8 Hz,
1H, 8); 6.87 (d, J = 8.1 Hz, 1H, 5); 7.04 (dd, J = 8.1; 2.1 Hz, 1H, 6); 7.16
(d, J = 2.1 Hz, 1H, 2); 7.54 (ddd, J = 15.9; 6.8; 2.8 Hz, 1H, 7).
[8,9-13C2]3-(3,4-Dihydroxyphenyl)propionic Acid. 43.7 mg
(0.24 mmol) of [13C2]CaffA was reduced to 42.6 mg (0.23 mmol,
yield 96%) of [13C2]3,4-diOHPPA. δH [ppm] 2.53 (ddt, J = 128.1; 7.2;
7.2 Hz, 2H, 8); 2.76 (m, 2H, 7); 6.57 (dd, J = 8.0; 2.0 Hz, 1H, 6); 6.72 (d,
J = 8.0 Hz; 1H, 5); 6.73 (bs, 1H, 2).
[8,9-13C2]3-(4-Hydroxyphenyl)propionic Acid. 31.0 mg (0.19
mmol) of [13C2]p-CA was reduced to 30.0 mg (0.18 mmol, yield 95%)
of [13C2]4-OHPPA. δH [ppm] 2.54 (ddt, J = 128.1; 7.4; 7.4 Hz, 2H 8);
2.81 (m, 2H, 7); 6.74 (m, 2H, 3;5); 7.07 (m, 2H, 2;6).
[8,9-13C2]3-(3-Hydroxyphenyl)propionic Acid. 40.2 mg (0.24
mmol) of [13C2]3-hydroxycinnamic acid was reduced to 39.6 mg (0.23
mmol, yield 96%) of [13C2]3-OHPPA. δH [ppm] 2.58 (ddt, J = 128.3;
7.3; 7.3 Hz, 2H, 8); 2.84 (m, 2H, 7); 6.70 (m, 3H, 2;4;6); 7.09 (t, J = 7.8
Hz, 1H; 5).
[8,9-13C2]3-Phenylpropionic Acid. 22.4 mg (0.15 mmol) of
[13C2]CinnA was reduced to 20.6 mg (0.14 mmol, yield 91%) of
[13C2]PPA. δH [ppm] 2.61 (ddt, J = 128.3, 7.4, 7.4 Hz, 2H, 8); 2.91 (m,
2H, 7); 7.18 (m, 1H, 4); 7.27 (m, 4H, 2;3;5;6).
[7,7-D2]Phenylacetic Acid. PPA (400 mg, 2.34 mmol), D2O (884
μL), and NaOD-solution (294 μL, 40 wt %) were placed in a high-
pressure bottle and air was exchanged for N2. After stirring overnight at
100 °C, the solution was cooled down to room temperature, acidified
with concentrated HCl (130 μL), and extracted into ethyl acetate. The
combined extracts were dried under reduced pressure. This procedure
was repeated twice to yield 303 mg of product (yield 75%) with 96% 2-
fold deuterium incorporation in benzylic position.29 δH [ppm] 7.25 (m,
1H, 4); 7.32 (m, 4H, 2;3;5;6).
[3,5-D2]4-Hydroxybenzoic Acid and [2,4,6-D3]3-Hydroxyben-
zoic Acid. 4-OHBA (100 mg, 0.725 mmol) and 3-OHBA (100 mg,
0.725 mmol), respectively, were placed in high-pressure bottles. D2O
(745 μL) and DCl-solution (124 μL, 35 wt %) were added, and the
mixtures were stirred for 48 h at 100 °C. Then, the solution was cooled
to 0 °C, and the crystallization product was collected by centrifugation,
washed with chilled water and lyophilized overnight. This procedure was
repeated once for [D2]4-OHBA and three times for [D3]3-OHBA.
Yields were 88.9 mg (89%) with 99% 2-fold deuterium incorporation for
4-OHBA and 84.8 mg (85%) with 98% 3-fold deuterium incorporation
for 3-OHBA, respectively. [D2]4-OHBA, δH [ppm] 7.92 (s, 2H, 2;6);
9.13 (bs, 1H, 4-OH); 3-OHBA, δH [ppm] 7.32 (s, 1H, 5).
C
DOI: 10.1021/acs.jafc.6b03318
J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry
Article
a
Table 1. Parameters of the Multiple Reaction Monitoring Using Specific Time Frames for the Monitored Transitions
analyte
3-OHHA
mass transition [m/z]
cone voltage [V]
collision energy voltage [V]
time frame [min]
194 → 150
196 → 151
137 → 93
139 → 95
178 → 134
180 → 135
24
11
5−16
[13C2]3-OHHA
4-OHBA
23
24
24
12
11
12
16−18.9
18.3−21
18.9−22
[D2]4-OHBA
HA
[13C2]HA
b
3,4-diOHPPA
181 → 137
c
181 → 109
b
[13C2]3,4-diOHPPA
183 → 138
c
183 → 109
3-OHBA
137 → 93
140 → 96
179 → 135
181 → 136
20
20
27
13
15
13
22−25
25−28
27−31
[D3]3-OHBA
CaffA
[13C2]CaffA
4-OHPPA
b
165 → 93
c
165 → 121
b
[13C2]4-OHPPA
diHFA
167 → 93
c
167 → 122
b
195 → 136
25
24
17
12
31−34.5
31−34.5
c
195 → 121
b
[13C2]diHFA
3-OHPPA
197 → 137
c
197 → 121
b
165 → 121
c
165 → 119
b
[13C2]3-OHPPA
167 → 122
c
167 → 120
p-CA
163 → 119
165 → 120
135 → 91
137 → 93
121 → 77
123 → 79
20
15
20
22
14
7
33.7−36.5
36.5−39.5
36.5−39.5
39−50
[13C2]p-CA
PAA
[D2]PAA
BA
11
15
[D2]BA
FA
b
193 → 134
c
193 → 178
b
[13C2]FA
195 → 135
c
195 → 180
PPA
149 → 105
151 → 106
147 → 103
149 → 104
24
22
11
11
47−54.4
54.2−65
[13C2]PPA
CinnA
[13C2]CinnA
a
BA, benzoic acid; CaffA, caffeic acid; CinnA, cinnamic acid; diHFA, dihydroferulic acid; 3,4-diOHPPA, 3-(3,4-dihydroxyphenyl)propionic acid; FA,
ferulic acid; HA, hippuric acid; 3-OHBA, 3-hydroxybenzoic acid; 4-OHBA, 4-hydroxybenzoic acid; 3-OHHA, 3-hydroxyhippuric acid; 3-OHPPA, 3-
(3-hydroxyphenyl)propionic acid; 4-OHPPA, 3-(4-hydroxyphenyl)propionic acid; PAA, phenylacetic acid; PPA, 3-phenylpropionic acid; p-CA, p-
b
c
coumaric acid. Quantifier mass transition. Qualifier mass transition.
[1,2-13C2]Hippuric Acid. [13C2]Glycine (100 mg, 1.27 mmol) was
dissolved in NaOH-solution (2 mL, 2 M). The solution was cooled to 0
°C, and benzoyl chloride (186 μL, 1.62 mmol) was added dropwise
while stirring vigorously. After stirring for another 2 h at 0 °C, the
solution was acidified to pH 1 and the crystallization product was filtered
under vacuum and dried afterward. The resulting byproduct BA was
dissolved in toluene (20 mL) at room temperature and removed by
vacuum filtration. Recrystallization in H2O yielded 195.5 mg (yield
85%) of the desired product. δH [ppm] 4.09 (dd, J = 139.2; 5.9 Hz, 2H,
CH2); 7.45 (m, 2H, 3;5); 7.52 (m, 1H, 4); 7.91 (m, 2H, 2;6).
mmol) was added slowly. The reaction mixture was stirred for 2 h,
before the solvent was evaporated under N2. The brown, solid residue
was suspended in toluene (400 μL) by ultrasonication and added
dropwise to a solution of glycine (54.4 mg, 0.725 mmol) in aqueous
NaOH (500 μL, 2 M) at 0 °C. After stirring for 1 h, the reaction mixture
was warmed to room temperature and stirred for another 4 h. Toluene
was evaporated under N2, the aqueous layer was acidified with
concentrated HCl and extracted with ethyl acetate three times. The
combined extracts were dried under reduced pressure to give an oily
residue. To remove remaining trimethylsilyl groups, the crude product
was dissolved in 2.5 mL of acetic acid/H2O/acetonitrile (3:1:1, v/v/v)
and kept at 60 °C for 16 h. After drying under reduced pressure, the
residue was dissolved in acetonitrile/H2O (1:1, v/v) and purified by
preparative HPLC, using a Luna C18(2) reversed phase column
(Phenomenex, 250 × 15 mm, 5 μm, 100 Å). Eluent A was 0.1% formic
acid in H2O, and eluent B was acetonitrile. The linear gradient (flow rate
of 8 mL min−1) increased from 10% B to 20% B within 15 min, followed
3-Hydroxyhippuric Acid and [1,2-13C2]3-Hydroxyhippuric
Acid. 3-OHBA (100 mg, 0.725 mmol), BSTFA (1 mL, 0.96 g, 3.73
mmol), and triethylamine (100 μL, 73 mg, 0.72 mmol) were allowed to
react in a 1.5 mL glass vial at 60 °C for 16 h. The solvent was removed
under reduced pressure at 40 °C, and the residue was lyophilized
afterward. Acetone (500 μL) was added to the oily product, the solution
was cooled to 0 °C, and thionyl chloride (52.6 μL, 86.28 mg, 0.725
D
DOI: 10.1021/acs.jafc.6b03318
J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry
Article
by rinsing and equilibration steps. Detection was carried out at 288 nm,
and collection of the desired compound was performed manually.
Evaporation of the eluent resulted in pure 3-hydroxyhippuric acid (3-
OHHA, 21.1 mg, 0,108 mmol, yield 15%). δH [ppm] 4.12 (d, J = 5.5 Hz,
2H, CH2); 7.00 (m, 1H, 4); 7.28 (t, J = 7.8 Hz, 1H, 5); 7.39 (m, 2H, 1;6).
The analogous synthesis using [13C2]glycine (55.8 mg, 0.725 mmol)
resulted in [13C2]3-OHHA (40.5 mg, 0,205 mmol, yield 28%). δH [ppm]
4.13 (ddd, J = 139.2; 5.8; 5.0 Hz, 1H, CH2); 7.00 (m, 1H, 4); 7.28 (t, J =
7.8 Hz, 1H, 5); 7.39 (m, 2H, 1;6).
Electrophilic aromatic substitution has been used for the
replacement of aromatic hydrogen atoms earlier. Using protocols
comparable to the method described by Kirby and Ogunkoya,30
H/D-exchange reactions were performed for 3,4-dihydroxyben-
zaldehyde, 4-hydroxybenzaldehyde, CaffA, p-CA, and 3,4-
diOHPPA. Although benzaldehydes were not required as
standard compounds in this project, they can be converted
into cinnamic acids or phenylpropionic acids as shown in Figure
2. However, electrophilic aromatic substitution under the
conditions tested did not result in the required 2-fold labeled
products. Either the replacement rate was marginal, or more than
two deuterium labels were incorporated into the molecule,
which, however, was not complete, resulting in a mixture of
differently labeled compounds.
Reduction of the particular cinnamic acids with D2 using
palladium (10%) on carbon as catalyst was another approach to
obtain 2-fold deuterium-labeled 3-phenylpropionic acids.31−33
Different from our expectations, it was, however, not possible to
exclusively incorporate two deuterium labels into the products.
Depending on the solvent used (methanol or methanol-d4), less
or more than two labels, respectively, were detected in the
products. This indicates that additional H/D-exchange mecha-
nisms between the gas atmosphere, solvent, and reactants were
triggered by the catalyst. Several mechanisms are described in the
literature aiming to explain similar observations,34−37 but these
were not further investigated for our compounds. Instead,
cinnamic acids were synthesized with two 13C-labels and partly
reduced to phenylpropionic acids.
Generally, cinnamic acids are obtained from benzaldehydes by
three different reaction types, all of which can be used to
incorporate two 13C-labels. In the Perkin-reaction, acetic
anhydride is used as solvent and reactant, and pyridine serves
as catalyst.38−40 The Horner−Wadsworth−Emmons-modifica-
tion of the Wittig-reaction uses trialkyl phosphonoacetates in
combination with anhydrous bases,41,42 and the Knoevenagel-
reaction with Doebner-modification uses malonic acid in
pyridine as reagent/solvent and piperidine or aniline as catalyst
to obtain cinnamic acids.43−45 Here, the Knoevenagel-reaction
with Doebner-modification was used, as it turned out to be the
least expensive and most practical reaction pathway. 2-Fold 13C-
labeled cinnamic acids were obtained, and hydrogenation with
H2 and palladium (10%) on carbon as catalyst resulted in 2-fold
13C-labeled 3-phenylpropionic acids (Figure 2).
HA can be obtained in a one-pot Schotten−Baumann
reaction46,47 using benzoyl chloride and glycine. Analogous
synthesis of 3-OHHA requires, however, protection of the
hydroxyl group of 3-OHBA; otherwise, intermolecular coupling
occurs after conversion to 3-hydroxybenzoyl chloride. Here,
trimethylsilylation was a suitable protection strategy, because the
carboxylic group still reacts to form benzoyl chloride, and the
protected hydroxyl group is stable enough to endure alkaline
Schotten−Baumann conditions. Deprotection is easily achieved
under acidic conditions (Figure 3).
Sample Analysis. Sample Preparation. A 15 g aliquot of vigorously
shaken milk was weighed into a 50 mL polypropylene vial. After adding
100 μL of a mixture of isotopically labeled standard compounds in
methanol/H2O (1:1, v/v) the sample was equilibrated for a minimum of
20 min. To deconjugate sulfonated and glucuronated analytes, the pH of
the sample was lowered with HCl (600 μL, 1 M) to pH 4.9, and sulfatase
(containing glucuronidase activity) from Helix pomatia (Sigma-Aldrich,
type H-1, 100 U) in acetate buffer (500 μL, 1 M, pH 4.9) was added. The
vial was incubated for 18 h at 38 °C with gentle agitation. Afterward, the
solution was acidified with HCl (2.1 mL, 1 M) to pH 1.5. To determine
the nonconjugated analytes only, the sample was directly acidified with
HCl (2.7 mL, 1 M) without prior addition of and incubation with the
enzymes. Then, acetonitrile (10 mL) was added followed by rigorous
shaking. Addition of MgSO4 (7.5 g) and NaCl (2 g) induced phase
separation. Samples were mixed rigorously for 3 × 30 s. After
centrifugation at 4000 rcf for 5 min, about 8 mL of the organic layer
were transferred into a 15 mL polypropylene vial, and MgSO4 (2 g) was
added. The sample was shaken rigorously for 3 × 30 s and centrifuged at
4000 rcf for 5 min. About 5 mL of the organic layer were transferred into
a 15 mL polypropylene vial. The residue after vacuum evaporation at 45
°C for 3.5 h was dissolved in methanol:H2O:concentrated HCl
(250:50:1, v/v/v) by ultrasonication for 5 min. This solution was
centrifuged during which phase separation partially occurred. Thus,
depending on the behavior of the individual sample, either the solution
or the upper layer (if phase separation occurred) was analyzed with LC-
MS/MS, directly and following 1:100 dilution with methanol:H2O
(50:50, v/v).
LC-MS/MS Analysis. HPLC was carried out using a phenyl hexyl
column (Phenomenex, Kinetex, 150 × 4.6 mm, 2.6 μm, 100 Å) with the
eluents 0.01% formic acid in H2O (A), 0.01% formic acid in methanol
(B), and 0.01% formic acid in acetonitrile (C). Elution at 24 °C was
carried out at a flow rate of 0.5 mL min−1 using a linear gradient: initial
conditions 95% A, 5% B, linear increase to 83% A, 17% B within 10 min,
held for 10 min, linear change to 87% A, 13% C within 1 min, held for 11
min, linear change to 75% A, 25% B within 1 min, held for 12 min, linear
increase to 100% B within 10 min, rinsing and equilibration steps. The
following MS parameters were used: ionization, ESI negative mode;
capillary, 3.4 kV; cone, analyte-dependent (Table 1); extractor, 3 V; RF
lens, 0.1 V; source temp, 120 °C; desolvation temperature, 350 °C;
desolvation gas flow, 750 L/h; cone gas flow, 50 L/h; LM resolution 1,
15; HM resolution 1, 15; ion energy 1, 0.8; entrance, −1; collision
energy, analyte-dependent (Table 1), exit, 1; LM resolution 2, 15; HM
resolution 2, 15; ion energy 2, 1.0; multiplier, 650; cell gas pirani, 3.2e−3.
Mass transitions and time frames for the multiple reaction monitoring-
modus are given in Table 1.
RESULTS AND DISCUSSION
■
Synthesis of Stable Isotope Labeled Standard Com-
pounds. Because isotopologues of most of the analytes (Figure
1) were either not commercially available or available at
exceptionally high prices only, synthetic routes were established
in our laboratory. Deuterium labeling is often the favored
pathway, because deuteration reagents are more affordable than
13C-labeled educts. On the other hand, 13C-labels are generally
more stable, and isotopic shifts during chromatographic
separation procedures (different retention times for isotopo-
logues) are less distinct for 13C-labeled compounds compared to
deuterium labeled compounds.
Synthetic routes to incorporate 13C-labels into benzoic acids
and PAA are not as trivial as for cinnamic acids or hippuric acids.
Therefore, H/D-exchanges were accomplished using harsher
conditions. For PAA, alkaline conditions facilitate keto−enol-
tautomerism, leading to a substitution of α-protons. For 3-
OHBA and 4-OHBA, electrophilic aromatic substitution in
aqueous deuterium chloride is facilitated by the hydroxyl groups,
which accelerate the reaction and direct substitution to both
ortho and para positions. Hence, 3-OHBA was the only analyte
which was not obtained as 2-fold labeled standard compound but
E
DOI: 10.1021/acs.jafc.6b03318
J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry
Article
dispersive solid phase extraction with primary and secondary
amines was omitted, because this cleanup step eliminates
carboxylic acids. Third, a larger aliquot of the extraction solution
was sampled and concentrated before injection to increase
method sensitivity by a factor of about 25. Although the
concentration step is most useful to analyze the minor
metabolites, it takes an additional 3.5 h. Fourth, enzyme
(sulfatase, glucuronidase) treatment enables the determination
of sulfate- and glucuronide-conjugated metabolites next to the
nonconjugated analytes. Because these conjugates can be formed
during hepatic phase II metabolism, it is advised to include them
into the method. Quantitative release of bound analytes was
verified by incubation of milk samples with different enzyme
activities (50 U, 100 U, and 300 U) in triplicate. For 10 out of 15
analytes, including the more dominant HA, BA, 3-OHHA, PAA,
and FA, no significant differences (p < 0.05) were found for the
higher enzyme concentrations (100 U and 300 U). Differences
were only found for p-CA and CinnA; however, this is more likely
due to their low concentrations close to the LOD. CaffA, 3,4-
diOHPPA, and 4-OHPPA were not quantifiable in these samples
(see below).
In some extracts, phase separation occurred following the
addition of methanol:H2O:concentrated HCl (250:50:1, v/v/v)
to the residue after concentrating an aliquot of the acetonitrile
extraction solution. Because of the acidic conditions, it was
assumed that the analytes are located in the organic fraction after
centrifugation. Analysis of the methanol fraction by LC-MS/MS
analysis confirmed this assumption.
LC-MS/MS Analysis. Suitable MS/MS mass transitions were
determined for all analytes (Table 1). If possible, a qualifier mass
transition was chosen in addition to the quantifier mass
transition. Chromatographic separation was achieved on a
core−shell phenyl−hexyl column using a ternary gradient. In
Figure 3. Synthesis of 3-hydroxyhippuric acid (3-OHHA) and [13C2]3-
OHHA starting with 3-hydroxybenzaldehyde (3-OHBA). I; N,O-
bis(trimethylsilyl)trifluoroacetamide, triethylamine, 60 °C, 16 h. II;
thionyl chloride. III; toluol, glycine, or [13C2]glycine in aqueous NaOH.
Asterisks indicate 13C-labels. TMS, trimethylsilyl.
contains three labels. Different from the benzaldehydes and the
cinnamic acids described above, 3-fold substitution was,
however, complete. Lack of an aromatic hydroxyl group prevents
a selective exchange in case of BA, but an appropriately labeled
isotopologue was commercially available.
Sample Preparation. The QuEChERS method, first
published in 2003,25 describes a frequently used procedure to
analyze pesticides in low-fat food products. Since its first
publication, multiple variations of the procedure expanded the
use of the QuEChERS method to a wide range of applications.48
Here, the principles of a previously described QuEChERS-based
method to extract pesticides from milk samples were used.26
Some modifications were, however, necessary to adjust the
procedure to the analytes of interest. First, the pH of the milk
samples was adjusted to lower values in order to optimize
extraction of our acidic analytes into acetonitrile. Second,
Figure 4. HPLC-MS/MS standard (A) and sample (B, organic cow milk, 3.8% fat) chromatograms combining chromatograms from different mass
transitions in single chromatograms. 3-Hydroxyhippuric acid (3-OHHA) and hippuric acid (HA) were analyzed after 1:100 dilution of the evaporated
and redissolved extract. BA, benzoic acid; CaffA, caffeic acid; CinnA, cinnamic acid; diHFA, dihydroferulic acid; 3,4-diOHPPA, 3-(3,4-
dihydroxyphenyl)propionic acid; FA, ferulic acid; 3-OHBA, 3-hydroxybenzoic acid; 4-OHBA, 4-hydroxybenzoic acid; 3-OHPPA, 3-(3-
hydroxyphenyl)propionic acid; 4-OHPPA, 3-(4-hydroxyphenyl)propionic acid; PAA, phenylacetic acid; PPA, 3-phenylpropionic acid; p-CA, p-
coumaric acid.
F
DOI: 10.1021/acs.jafc.6b03318
J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry
Article
b
Table 2. Lowest Calibration Levels (LCL), Calibration Equations, and Coefficients of Determination (R2) for the Analytes and
a
Their Corresponding Isotopically Labeled Standard Compounds
LCL [μg/mL]
calibration equation (LCL − 1 μg/mL)
R2
calibration equation (1 μg/mL − 11 μg/mL)
R2
3-OHHA
[13C2]3-OHHA
4-OHBA
0.005
0.005
0.025
0.025
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.025
0.025
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.025
0.025
0.005
0.005
0.025
0.025
0.005
0.005
y = −6163.66x2 + 54610.42x + 3.20
y = −5129.54 x2 + 48533.708 x - 1.86
y = −3652.18 x2 + 64510.75 x + 29.31
y = −7741.37x2 + 69114.40x − 68.40
y = −4796.57x2 + 41409.25x + 5.96
y = −5355.09x2 + 41557.83x + 0.88
y = −453.56x2 + 27213.196x − 17.70
y = −464.67x2 + 24370.40x − 11.35
y = −4517.41x2 + 56068.35x − 0.81
y = −6873.74x2 + 65166.19x + 13.61
y = −1527.42x2 + 77767.75x − 64.29
y = −2702.00x2 + 91859.04x − 42.57
y = 68.39x2 + 1943.31x −1.36
0.99999
0.99999
0.99995
0.99939
0.99996
0.99985
0.99992
0.99985
0.99982
0.99989
0.99997
0.99986
0.99941
0.99982
0.99962
0.99974
0.99997
0.99998
0.99971
0.99988
0.99982
0.99959
0.99992
0.99984
0.99909
0.99870
0.99997
0.99994
0.99998
0.99989
y = −1240.58x2 + 40879.31x + 8794.48
y = −1024.93x2 + 35527.96x + 8935.23
y = −1560.89x2 + 56804.96x + 6549.16
y = −1533.40x2 + 57405.16x + 7849.06
y = −704.16x2 + 32078.89x + 4934.99
y = −619.48x2 + 31072.73x + 6224.93
y = −613.59x2 + 24708.4546x + 2737.89
y = −434.04x2 + 20663.44x + 3842.84
y = −930.82x2 + 45440.91x + 7283.44
y = −1415.59x2 + 55038.14x + 5370.07
y = −1791.70x2 + 67860.64x + 13119.72
y = −2008.78x2 + 78871.14x + 15054.20
y = −35.83x2 + 2106.87x + 12.93
0.99969
0.99983
0.99993
0.99995
0.99974
0.99951
0.99967
0.99948
0.99958
0.99994
0.99914
0.99932
0.99988
0.99972
0.99854
0.99863
0.99925
0.99945
0.99988
0.99993
0.99982
0.99999
0.99984
0.99989
0.99987
0.99993
0.99999
1.00000
0.99999
0.99999
[D2]4-OHBA
HA
[13C2]HA
3,4-diHOPPA
[13C2]3,4-diHOPPA
3-OHBA
[D3]3-OHBA
CaffA
[13C2]CaffA
4-OHPPA
[13C2]4-OHPPA
diHFA
y = 108.86x2 + 1834.36x − 0.74
y = −30.76x2 + 1947.76x + 75.65
y = −7749.06x2 + 52160.46x − 16.82
y = −7915.49x2 + 51371.02x − 20.6
y = −7187.23x2 + 87843.14x − 13.11
y = −7048.66x2 + 95509.69x − 11.98
y = −1688.11x2 + 128469.96x − 14.23
y = −2681.35x2 + 140329.39x − 32.94
y = 109.72x2 + 14030.99x + 0.06
y = −178.86x2 + 13732.97x − 7.06
y = −92.43x2 + 4744.17x + 86.33
y = −262.17x2 + 4375.13x − 3.39
y = 1284.21x2 + 46816.70x + 5.14
y = 2273.74x2 + 45554.03x + 0.97
y = 291.68x2 + 2477.92x + 0.21
y = −683.80x2 + 30437.26x + 17065.50
y = −698.08x2 + 29930.07x + 15733.76
y = −1453.87x2 + 66017.93x + 19445.92
y = −1709.90x2 + 73005.74x + 18985.64
y = −2185.78x2 + 103535.68x + 25990.06
y = −2291.10x2 + 112070.76x + 28014.01
y = −59.73x2 + 16607.21 × −2219.39
y = −26.59x2 + 15537.15x −1653.99
y = −37.87x2 + 4654.97x + 175.15
y = −27.58x2 + 4126.50x + 83.88
[13C2]diHFA
3-OHPPA
[13C2]3-OHPPA
p-CA
[13C2]p-CA
PAA
[D2]PAA
BA
[D2]BA
FA
[13C2]FA
y = −536.75x2 + 46297.26x + 2725.98
y = −459.34x2 + 44581.16x + 4263.07
y = 11.68x2 + 2961.81x - 210.83
PPA
[13C2]PPA
CinnA
y = 271.99x2 + 2736.43x − 2.20
y = 17.21x2 + 3182.75x - 194.23
y = 404.37x2 + 14060.01x + 8.96
y = 439.62x2 + 15382.77x + 4.81
y = −87.45x2 + 15405.87x - 865.67
[13C2]CinnA
y = −68.05x2 + 16382.18x - 424.01
a
BA, benzoic acid; CaffA, caffeic acid; CinnA, cinnamic acid; diHFA, dihydroferulic acid; 3,4-diOHPPA, 3-(3,4-dihydroxyphenyl)propionic acid; FA,
ferulic acid; HA, hippuric acid; 3-OHBA, 3-hydroxybenzoic acid; 4-OHBA, 4-hydroxybenzoic acid; 3-OHHA, 3-hydroxyhippuric acid; 3-OHPPA, 3-
(3-hydroxyphenyl)propionic acid; 4-OHPPA, 3-(4-hydroxyphenyl)propionic acid; PAA, phenylacetic acid; PPA, 3-phenylpropionic acid; p-CA, p-
b
coumaric acid. y = peak area, x = analyte concentration [μg/mL].
Figure 4, the chromatograms from all quantifier mass transitions
of all analytes of a milk sample (organic cow milk containing
3.8% fat) are combined in a single chromatogram (data from 3-
OHHA and HA are taken from the analysis of the diluted sample;
see below) and compared to a chromatogram of standard
compounds. The chromatographic separation of 4-OHBA and
3,4-diOHPPA is not affected by the high concentration of
coeluting HA, which shows a 2 min broad, tailing peak in the
concentrated sample. A maximum of eight mass transitions is
recorded within a given time frame (Table 1), resulting in
sufficient (at least 12) data points per peak to ensure correct peak
integration. For some quantifier mass transitions, more than one
peak appears in the chromatogram. In these cases, retention
times of analytes and isotopically labeled standard compounds
need to be compared, and, where defined, qualifier mass
transitions have to be considered for an unambiguous
identification.
The first quantitative experiments were performed to roughly
estimate the contents of all analytes in different milk samples in
order to add comparable amounts of isotopically labeled
standard compounds to the milk samples. Due to large
differences among the analyte concentrations in milk samples,
two measurements (concentrated and in 1:100 dilution) of each
sample are advised. Without dilution, HA concentrations exceed
the calibration range; also, the 3-OHHA levels are high and
should therefore be assessed from measuring the diluted sample
extract.
Method Validation and Application. The lowest calibra-
tion levels (LCL) were determined by diluting mixtures of all
standard compounds until signal-to-noise ratios of at least 9:1
(LCL, Table 2) were reached. Calibration ranges were chosen
from the respective LCL up to concentrations of 11 μg mL−1,
divided into two sections: section 1, LCL to 1 μg mL−1; and
section 2, 1 μg mL−1 to 11 μg mL−1. Both sections covered six
equidistant calibration points. Each calibration point was
measured in triplicate, and, because homogeneity of variances
was not given, weighted regression (with reduced chi2, weighting
factor 1/standard deviation2) was used. The obtained data fit
second order polynomial regression better than linear regression,
which was also confirmed by residual analysis. Calibration
equations and the corresponding coefficients of determination
(R2) are given in Table 2. Assessing both limit of detection
(LOD) and limit of quantitation (LOQ) is not a straightforward
process, because empty matrices (milk samples which do not
contain our analytes) are not available and mimicking a milk
matrix is problematic because both caseins and whey protein may
adsorb phenolic compounds, thus potentially carrying analytes
into our empty matrix. Spiking milk samples with our isotopically
G
DOI: 10.1021/acs.jafc.6b03318
J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry
Article
Table 3. Limits of Detection (LOD), Limits of Quantitation (LOQ), and Recovery Rates of Isotopically Labeled Analytes Added to
a
Conventional or Organic Milk Samples with Different Fat Contents (1.5; 3.5; or 3.8%)
recovery rate in
recovery rate in
LOD in conventional
milk 3.5% fat [μg/kg]
LOQ in conventional
milk 3.5% fat [μg/kg]
conventional milk 1.5% fat conventional milk 3.5% fat recovery rate in organic
analyte
[%]
[%]
milk 3.8% fat [%]
[13C2]3-OHHA
[D2]4-OHBA
[13C2]HA
[13C2]3,4-diOHPPA
[D3]3-OHBA
[13C2]CaffA
[13C2]4-OHPPA
[13C2]diHFA
[13C2]3-OHPPA
[13C2]p-CA
0.15
0.2
0.4
0.7
67
52
61
41
60
76
43
22
29
60
31
39
90
59
60
69
56
64
37
63
74
41
21
25
64
30
44
94
61
62
68
55
63
38
59
84
30
47
45
64
29
42
90
47
56
b
b
N/A
N/A
0.7
0.15
0.15
3
2
0.4
0.4
9
0.2
0.1
0.03
0.7
2
0.7
0.3
0.1
2
[D2]PAA
[D2]BA
7
[13C2]FA
0.1
0.2
0.07
0.3
0.6
0.2
[13C2]PPA
[13C2]CinnA
a
BA, benzoic acid; CaffA, caffeic acid; CinnA, cinnamic acid; diHFA, dihydroferulic acid; 3,4-diOHPPA, 3-(3,4-dihydroxyphenyl)propionic acid; FA,
ferulic acid; HA, hippuric acid; 3-OHBA, 3-hydroxybenzoic acid; 4-OHBA, 4-hydroxybenzoic acid; 3-OHHA, 3-hydroxyhippuric acid; 3-OHPPA, 3-
(3-hydroxyphenyl)propionic acid; 4-OHPPA, 3-(4-hydroxyphenyl)propionic acid; PAA, phenylacetic acid; PPA, 3-phenylpropionic acid; p-CA, p-
b
coumaric acid. See text.
labeled compounds is an option; however, due to the small but
existing spectral overlap between the labeled and unlabeled
analytes, this strategy does not work for analytes with large
concentrations in the milk samples. Nevertheless, this approach
was used to determine approximate LODs and LOQs, and
mixtures of all isotopically labeled standard compounds were
added in decreasing concentrations to samples of 15 g of
conventional milk with 3.5% fat. Samples were processed as
described with enzyme treatment. Concentrations, which
resulted in a signal-to-noise-ratio of 3:1 and 9:1, were defined
as LOD and LOQ, respectively (Table 3). However, the LOD
and LOQ of [13C2]HA are not assessable because the high levels
of naturally occurring HA result in a severe overloading of the
HPLC column and, therefore, in a broad signal with shifted
retention time. Additionally, 0.1% of the natural HA are recorded
with the same mass transition as for [13C2]HA because of the
natural abundance of 13C as described above.
The applicability of the method was tested by analyzing three
different milk samples (conventional cow milk, 1.5% fat;
conventional cow milk, 3.5% fat; organic cow milk, 3.8% fat) in
triplicate. As shown in Figure 5, it was possible to quantitate all
analytes with the exception of 3,4-diOHPPA, 4-OHPPA, and
nonconjugated CaffA in these samples. Incubation with sulfatase
and glucuronidase before extraction resulted in similar or higher
concentrations of all analytes compared to nonenzyme treatment
data. Also, metabolites that can only be ester-conjugated (BA,
PPA, PAA, and CinnA) were determined in higher concen-
trations after enzyme incubation.
these values only influence the method sensitivity, but should not
affect the accuracy of the method. To verify this assumption,
spiking procedures were performed in triplicate by adding a
mixture of nonlabeled standard compounds (together with the
labeled standards) to conventional milk samples containing 3.5%
fat. Recovery rates of 90−110% were determined for most of the
analytes. However, recovery rates of 4-OHBA, 3-PPA, and
CinnA were only around 50−70%. Therefore, data for these
analytes should be labeled as semiquantitative only.
Comparing the phenolic acid contents in the three analyzed
milk samples, promising differences in the metabolite patterns
can be observed. The concentrations of 3-OHPPA, 3-OHBA, 3-
OHHA, 4-OHBA, and BA are higher in the analyzed organic milk
sample compared to the milk samples from conventional
production, whereas diHFA and PAA seem to be more abundant
in conventional milk samples. However, whether or not these
differences are only due to the amounts of ingested FA or
whether and how other ingested phenolic compounds influence
these patterns cannot be answered at this point. Larger studies
with milk samples of known history will be necessary to estimate
the potential of differentiating organic and conventional milk
based on ferulic acid metabolites. In comparison to the results
from other studies,20 some analyte concentrations differ widely.
However, Besle et al. showed that well-defined diets result in
variations of the phenolic acid contents in milk.20
In conclusion, a stable isotope dilution analysis approach based
on HPLC-MS/MS was developed for sensitive, accurate, and
precise determination of 15 phenolic acids (nonconjugated and
conjugated) in cow’s milk. One essential requirement, however,
is the appropriate addition of labeled standard compounds to the
samples, particularly when second-order calibration curves are
used. Although only three milk samples were analyzed by using
this method so far, differences among the milk samples profiles
suggest that the application of the developed method on a large
number of samples of known origin in combination with
multivariate data analysis may help to trace back cattle forages.
Standard deviations of the total concentrations (free and
conjugated) of the metabolites for the analyzed milk samples
representing the precision of the method range between <1% and
8.9%. The data were also used to calculate recovery rates of the
added isotopically labeled standard compounds. Considering
that after QuEChERS extraction only about 5 out of 10 mL of the
organic fraction was used for concentration and LC-MS/MS
analysis, recovery rates are between 21% for [13C2]diHFA and
94% for [13C2]FA (Table 3). However, although some recovery
rates are apparently quite low, it needs to be mentioned that
H
DOI: 10.1021/acs.jafc.6b03318
J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry
Article
Figure 5. Concentrations of phenolic acids in conventional and organic cow milk samples with different fat contents (1.5, 3.5, or 3.8%) by LC-MS/MS.
The sums of nonconjugated and conjugated analytes were measured after incubation with sulfatase and glucuronidase. Data are shown as mean
standard deviation, n = 3. BA, benzoic acid; CaffA, caffeic acid; CinnA, cinnamic acid; diHFA, dihydroferulic acid; 3,4-diOHPPA, 3-(3,4-
dihydroxyphenyl)propionic acid; FA, ferulic acid; HA, hippuric acid; 3-OHBA, 3-hydroxybenzoic acid; 4-OHBA, 4-hydroxybenzoic acid; 3-OHHA, 3-
hydroxyhippuric acid; 3-OHPPA, 3-(3-hydroxyphenyl)propionic acid; 4-OHPPA, 3-(4-hydroxyphenyl)propionic acid; PAA, phenylacetic acid; PPA, 3-
phenylpropionic acid; p-CA, p-coumaric acid.
Notes
AUTHOR INFORMATION
■
The authors declare no competing financial interest.
Corresponding Author
*Post address: Karlsruhe Institute of Technology (KIT),
Institute of Applied Biosciences, Department of Food Chemistry
and Phytochemistry, Adenauerring 20a, 76131 Karlsruhe,
Germany; Telephone: (+49) 721 608 42936; Fax: (+49) 721
608 47255; E-mail: mirko.bunzel@kit.edu.
ABBREVATIONS
■
BA, benzoic acid; BSTFA, N,O-bis(trimethylsilyl)-
trifluoroacetamide; CaffA, caffeic acid; CinnA, cinnamic acid;
diHFA, dihydroferulic acid; 3,4-diOHPPA, 3-(3,4-
I
DOI: 10.1021/acs.jafc.6b03318
J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry
Article
(20) Besle, J. M.; Viala, D.; Martin, B.; Pradel, P.; Meunier, B.;
dihydroxyphenyl)propionic acid; ESI, electrospray ionization;
FA, ferulic acid; HA, hippuric acid; MS, mass spectrometry;
NMR, nuclear magnetic resonance; 3-OHBA, 3-hydroxybenzoic
acid; 4-OHBA, 4-hydroxybenzoic acid; 3-OHHA, 3-hydrox-
yhippuric acid; 3-OHPPA, 3-(3-hydroxyphenyl)propionic acid;
4-OHPPA, 3-(4-hydroxyphenyl)propionic acid; PAA, phenyl-
acetic acid; PPA, 3-phenylpropionic acid; p-CA, p-coumaric acid
́
Berdague, J. L.; Fraisse, D.; Lamaison, J. L.; Couoln, J. B. Ultraviolet-
absorbing compounds in milk are related to forage polyphenols. J. Dairy
Sci. 2010, 93, 2846−2856.
(21) O’connell, J.; Fox, P. Significance and applications of phenolic
compounds in the production and quality of milk and dairy products: a
review. Int. Dairy J. 2001, 11, 103−120.
(22) Rechner, A. R.; Kuhnle, G.; Bremner, P.; Hubbard, G. P.; Moore,
K. P.; Rice-Evans, C. A. The metabolic fate of dietary polyphenols in
humans. Free Radical Biol. Med. 2002, 33, 220−235.
REFERENCES
■
(23) Clayton, T. A. Metabolic differences underlying two distinct rat
urinary phenotypes, a suggested role for gut microbial metabolism of
phenylalanine and a possible connection to autism. FEBS Lett. 2012,
586, 956−961.
(1) Hartley, R. D.; Ford, C. W. Phenolic constituents of plant cell walls
and wall biodegradability. ACS Symp. Ser. 1989, 399, 137−145.
(2) Eraso, F.; Hartley, R. D. Monomeric and dimeric phenolic
constituents of plant cell walls  possible factors influencing wall
biodegradability. J. Sci. Food Agric. 1990, 51, 163−170.
(3) Ishii, T. Structure and functions of feruloylated polysaccharides.
Plant Sci. 1997, 127, 111−127.
(24) Kusche, D.; Kuhnt, K.; Ruebesam, K.; Rohrer, C.; Nierop, A. F.;
Jahreis, G.; Baars, T. Fatty acid profiles and antioxidants of organic and
conventional milk from low-and high-input systems during outdoor
period. J. Sci. Food Agric. 2015, 95, 529−539.
(4) Colquhoun, I. J.; Ralet, M. C.; Thibault, J.-F.; Faulds, C. B.;
Williamson, G. Structure identification of feruloylated oligosaccharides
from sugar-beet pulp by NMR spectroscopy. Carbohydr. Res. 1994, 263,
243−256.
(5) Saulnier, L.; Marot, C.; Chanliaud, E.; Thibault, J.-F. Cell wall
polysaccharide interactions in maize bran. Carbohydr. Polym. 1995, 26,
279−287.
(6) Rombouts, F. M.; Thibault, J.-F. Feruloylated pectic substances
from sugar-beet pulp. Carbohydr. Res. 1986, 154, 177−187.
(7) Ralph, J.; Quideau, S.; Grabber, J. H.; Hatfield, R. D. Identification
and synthesis of new ferulic acid dehydrodimers present in grass cell
walls. J. Chem. Soc., Perkin Trans. 1 1994, 1, 3485−3498.
(8) Bunzel, M. Chemistry and occurrence of hydroxycinnamate
oligomers. Phytochem. Rev. 2010, 9, 47−64.
(9) Russell, W. R.; Scobbie, L.; Chesson, A.; Richardson, A. J.; Stewart,
C. S.; Duncan, S. H.; Drew, J. E.; Duthie, G. G. Anti-inflammatory
implications of the microbial transformation of dietary phenolic
compounds. Nutr. Cancer 2008, 60, 636−642.
(10) Gonthier, M. P.; Remesy, C.; Scalbert, A.; Cheynier, V.; Souquet,
J. M.; Poutanen, K.; Aura, A. M. Microbial metabolism of caffeic acid and
its esters chlorogenic and caftaric acids by human faecal microbiota in
vitro. Biomed. Pharmacother. 2006, 60, 536−540.
(11) Anson, N. M.; Selinheimo, E.; Havenaar, R.; Aura, A.-M.; Mattila,
I.; Lehtinen, P.; Bast, A.; Poutanen, K.; Haenen, G. R. M. M.
Bioprocessing of wheat bran improves in vitro bioaccessibility and
colonic metabolism of phenolic compounds. J. Agric. Food Chem. 2009,
57, 6148−6155.
̌
(25) Anastassiades, M.; Lehotay, S. J.; Stajnbaher, D.; Schenck, F. J.
Fast and easy multiresidue method employing acetonitrile extraction/
partitioning and “dispersive solid-phase extraction” for the determi-
nation of pesticide residues in produce. J. AOAC Int. 2003, 86, 412−431.
̌ ́
(26) Lehotay, S. J.; Mastovska, K.; Yun, S. J. Evaluation of two fast and
easy methods for pesticide residue analysis in fatty food matrixes. J.
AOAC Int. 2005, 88, 630−638.
(27) Gottlieb, H. E.; Kotlyar, V.; Nudelman, A. NMR chemical shifts of
common laboratory solvents as trace impurities. J. Org. Chem. 1997, 62,
7512−7515.
(28) Krings, U.; Pilawa, S.; Theobald, C.; Berger, R. Phenyl propenoic
side chain degradation of ferulic acid by Pycnoporus cinnabarinus 
elucidation of metabolic pathways using [5-2H]-ferulic acid. J. Biotechnol.
2001, 85, 305−314.
(29) Orfanopoulos, M.; Smonou, I.; Foote, C. S. Intermediates in the
ene reactions of singlet oxygen and N-phenyl-1,2,4-triazoline-3,5-dione
with olefins. J. Am. Chem. Soc. 1990, 112, 3607−3614.
(30) Kirby, G.; Ogunkoya, L. 1278. Deuterium and tritium exchange
reactions of phenols and the synthesis of labelled 3,4-dihydroxypheny-
lalanines. J. Chem. Soc. 1965, 6914−6922.
(31) Zhao, H.; Neamati, N.; Mazumder, A.; Sunder, S.; Pommier, Y.;
Burke, T. R. Arylamide inhibitors of HIV-1 integrase. J. Med. Chem.
1997, 40, 1186−1194.
(32) Moridani, M. Y.; Scobie, H.; O’Brien, P. J. Metabolism of caffeic
acid by isolated rat hepatocytes and subcellular fractions. Toxicol. Lett.
2002, 133, 141−151.
(33) Fumeaux, R.; Menozzi-Smarrito, C.; Stalmach, A.; Munari, C.;
Kraehenbuehl, K.; Steiling, H.; Crozier, A.; Williamson, G.; Barron, D.
First synthesis, characterization, and evidence for the presence of
hydroxycinnamic acid sulfate and glucuronide conjugates in human
biological fluids as a result of coffee consumption. Org. Biomol. Chem.
2010, 8, 5199−5211.
(34) Sajiki, H.; Kurita, T.; Esaki, H.; Aoki, F.; Maegawa, T.; Hirota, K.
Complete replacement of H2 by D2 via Pd/C-catalyzed H/D exchange
reaction. Org. Lett. 2004, 6, 3521−3523.
(35) Strathdee, G. G. Anchored homogeneous-type catalysts for H-D
exchange. US Patent Number 3900557, 1975.
(36) Sajiki, H.; Aoki, F.; Esaki, H.; Maegawa, T.; Hirota, K. Efficient
CH/CD exchange reaction on the alkyl side chain of aromatic
compounds using heterogeneous Pd/C in D2O. Org. Lett. 2004, 6,
1485−1487.
(37) Esaki, H.; Aoki, F.; Umemura, M.; Kato, M.; Maegawa, T.;
Monguchi, Y.; Sajiki, H. Efficient H/D exchange reactions of alkyl-
substituted benzene derivatives by means of the Pd/C−H2−D2O
system. Chem. - Eur. J. 2007, 13, 4052−4063.
(38) Perkin, W. H. XLVIII.On the action of acetic anhydride upon
the hydrides of salicyl, ethyl−salicyl, &c. J. Chem. Soc. 1867, 20, 586−
591.
(12) Rosazza, J. P. N.; Huang, Z.; Dostal, L.; Volm, T.; Rousseau, B.
Review: Biocatalytic transformations of ferulic acid: An abundant
aromatic natural product. J. Ind. Microbiol. 1995, 15, 457−471.
(13) Chesson, A.; Provan, G. J.; Russell, W. R.; Scobbie, L.; Richardson,
A. J.; Stewart, C. Hydroxycinnamic acids in the digestive tract of
livestock and humans. J. Sci. Food Agric. 1999, 79, 373−378.
(14) McSweeney, C.; Dulieu, A.; Bunch, R. Butyrivibrio spp. and other
xylanolytic microorganisms from the rumen have cinnamoyl esterase
activity. Anaerobe 1998, 4, 57−65.
(15) Jansen, G. A.; Wanders, R. J. Alpha-oxidation. Biochim. Biophys.
Acta, Mol. Cell Res. 2006, 1763, 1403−1412.
(16) Spencer, J. P.; Chowrimootoo, G.; Choudhury, R.; Debnam, E. S.;
Srai, S. K.; Rice-Evans, C. The small intestine can both absorb and
glucuronidate luminal flavonoids. FEBS Lett. 1999, 458, 224−230.
(17) Mao, L.-F.; Chu, C.; Schulz, H. Hepatic β-oxidation of 3-
phenylpropionic acid and the stereospecific dehydration of (R)-and (S)-
3-hydroxy-3-phenylpropionyl-CoA by different enoyl-CoA hydratases.
Biochemistry 1994, 33, 3320−3326.
(18) Zhao, Z.; Moghadasian, M. H. Chemistry, natural sources, dietary
intake and pharmacokinetic properties of ferulic acid: A review. Food
Chem. 2008, 109, 691−702.
(19) Zhao, Z.; Egashira, Y.; Sanada, H. Ferulic acid is quickly absorbed
from rat stomach as the free form and then conjugated mainly in liver. J.
Nutr. 2004, 134, 3083−3088.
(39) Bacharach, G.; Brogan, F. The action of pyridine as a catalyst in
Perkin’s synthesis of cinnamic acid. J. Am. Chem. Soc. 1928, 50, 3333−
3334.
J
DOI: 10.1021/acs.jafc.6b03318
J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry
Article
(40) Pawar, P. M.; Jarag, K. J.; Shankarling, G. S. Environmentally
benign and energy efficient methodology for condensation: an
interesting facet to the classical Perkin reaction. Green Chem. 2011,
13, 2130−2134.
(41) Horner, L.; Hoffmann, H.; Wippel, H. G.; Klahre, G.
Phosphororganische Verbindungen, XX. Phosphinoxyde als Olefinier-
ungsreagenzien. Chem. Ber. 1959, 92, 2499−2505.
(42) Wadsworth, W. S.; Emmons, W. D. The utility of phosphonate
carbanions in olefin synthesis. J. Am. Chem. Soc. 1961, 83, 1733−1738.
(43) Knoevenagel, E. Condensationen zwischen Malonester und
Aldehyden unter dem Einfluss von Ammoniak und organischen
Aminen. Ber. Dtsch. Chem. Ges. 1898, 31, 2585−2595.
(44) Doebner, O. Synthese der Sorbinsaure. Ber. Dtsch. Chem. Ges.
̈
1900, 33, 2140−2142.
(45) Robbins, R. J.; Schmidt, W. F. Optimized synthesis of four
isotopically labeled (13C-enriched) phenolic acids via a malonic acid
condensation. J. Labelled Compd. Radiopharm. 2004, 47, 797−806.
(46) Schotten, C. Ueber die Oxydation des Piperidins. Ber. Dtsch.
Chem. Ges. 1884, 17, 2544−2547.
(47) Baumann, E. Ueber eine einfache Methode der Darstellung von
Benzoesaureathern. Ber. Dtsch. Chem. Ges. 1886, 19, 3218−3222.
̈
̈
̈
(48) Rejczak, T.; Tuzimski, T. A review of recent developments and
trends in the QuEChERS sample preparation approach. Open Chem.
2015, 13.10.1515/chem-2015-0109
K
DOI: 10.1021/acs.jafc.6b03318
J. Agric. Food Chem. XXXX, XXX, XXX−XXX