RESEARCH ARTICLE
www.mnf-journal.com
Phenolic Metabolites in the Urine and Plasma of Healthy
Men After Acute Intake of Purple Potato Extract Rich in
Methoxysubstituted Monoacylated Anthocyanins
Johanna Jokioja,* Jasmine Percival, Mark Philo, Baoru Yang, Paul A. Kroon,
and Kaisa M. Linderborg
Scope: Structurally stable acylated anthocyanins have potential in various
food applications but the effects of acylation and methoxysubstitution on
anthocyanin metabolism are poorly understood. This is the first study
thoroughly investigating phenolic metabolites, their time-wise changes, and
pharmacokinetics following an acute intake of methoxysubstituted
monoacylated anthocyanins.
Methods and Results: Healthy male volunteers (n = 17) consumed a yellow
potato meal with and without purple potato extract rich in acylated
anthocyanins (152 mg) and hydroxycinnamic acid conjugates (140 mg).
Ultra-high performance liquid chromatography-tandem mass spectrometry
(UHPLC-MS/MS) is used for identification and quantification of metabolites
from serially collected urine and plasma. While the parent anthocyanins are
not detected, 28 phenolic metabolites from urine and 14 from plasma are
quantified, including hydroxybenzoic and hydroxycinnamic acids and
protocatechuic acid sulfates and glucuronides; three (catechol, gallic
acid-4-O-glucuronide, and 2-methoxybenzoic acid) are detected for the first
time after anthocyanin-rich food. Urinary hippuric acid is the most abundant
with an increase of 139 µM mM−1 creatinine after the treatment. A large
additional set of tentatively identified phenolic metabolites are detected. Late
urinary peak time values suggest colonic degradation.
Conclusion: Acylated anthocyanins are more bioavailable than earlier
reported after extensive degradation in human and/or colonial metabolism to
phenolic metabolites, which may be further conjugated and demethylated.
Dr. J. Jokioja, Prof. B. Yang, Prof. K. M. Linderborg
Food Chemistry and Food Development
Department of Life Technologies
University of Turku
Turku FIN-20014, Finland
E-mail: johanna.jokioja@utu.fi
J. Percival, M. Philo, Dr. P. A. Kroon
The Quadram Institute Bioscience
Norwich Research Park
Norwich NR4 7UQ, Norfolk, United Kingdom
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/10.1002/mnfr.202000898
© 2021 The Authors. Molecular Nutrition & Food Research published by
Wiley-VCH GmbH. This is an open access article under the terms of the
Creative Commons Attribution License, which permits use, distribution
and reproduction in any medium, provided the original work is properly
cited.
DOI: 10.1002/mnfr.202000898
1. Introduction
Anthocyanins, the red-purple col-
orants of berries and fruits, are hy-
poglycemic and decrease the risk of type
II diabetes.[1,2] While the health proper-
ties of dietary components are typically
linked to bioavailability, the bioavailabil-
ity of anthocyanins is extremely low and
affected by their vast structural diversity.
For example, the urinary recovery of pure
13C-labeled cyanidin-3-O-glucoside was
less than 2%,[3] and 0.03% for bilberry
anthocyanins.[4] Acylation of antho-
cyanins, i.e., linking one or more organic
acids to the glycosyl moiety, decreases
the bioavailability even further. The
mainly diacylated cyanidin diglucosides
of red cabbage are recovered four times
less, whereas the monoacylated cyanidin
xylosides of purple carrots are recov-
ered 14 times less than the nonacylated
anthocyanins of these foods.[5,6] Also,
acylated anthocyanins are acutely hypo-
glycemic as detected with both cooked
purple potatoes[7] and their extract[8]
(99% acylated anthocyanins). Therefore,
the health effects of anthocyanins may
be mediated by their more bioavailable
metabolites.[9,10]
Acylated anthocyanins have more po-
tential for food industrial applications
than nonacylated ones due to their enhanced structural
stability[11] and they are common in everyday diets as part of
pigmented vegetables and tubers.[5,6,8,12,13] However, their degra-
dation in humans has not been studied extensively previously.
Earlier clinical trials investigating the metabolites of antho-
cyanins have focused mainly on the pure common nonacy-
lated anthocyanin glycoside, cyanidin-3-O-glucoside,[3,14] foods
rich in nonacylated anthocyanins, such as various berries,[4,15–23]
and to some extent on foods with a mix of nonacylated
and acylated anthocyanins, such as grapes.[24,25] One study
demonstrated that an acute intake of purple potatoes rich
in acylated anthocyanins leads to the elevation of the total
concentration of phenolics in biofluids of healthy volunteers
(n= 5). Only a few of the phenolics were tentatively identified due
to the lack of standard compounds, and the study did not include
Mol. Nutr. Food Res. 2021, 65, 2000898 2000898 (1 of 11) © 2021 The Authors. Molecular Nutrition & Food Research published by Wiley-VCH GmbH
www.advancedsciencenews.com www.mnf-journal.com
a control arm.[26] Altogether, these papers suggest that in addition
to possibleminor absorption, nonacylated anthocyanins undergo
fission of the heterocyclic C-ring leading to phenolic metabo-
lites, such as hydroxybenzoic acids, phenyl acetic acids, phenyl
propionic acids, hydroxycinnamic acids, phenyl alcohols, phenyl
aldehydes, and hippuric acids. After entering enterohepatic cir-
culation, the metabolites may be further conjugated in the phase
II metabolism. The in vivo degradation of acylated anthocyanins,
however, is still not understood.
The aim of this study was to investigate the postprandial struc-
tural changes of the ingested purple potato extract (PPE) from
Solanum tuberosum L. ′Synkeä Sakari′ rich inmethoxysubstituted
anthocyanin glucosides with a rutinosyl moiety acylated to a caf-
feic, coumaric, or ferulic acid as our previous clinical trials gave
evidence that the potatoes and their extract decrease postprandial
glycemia and insulinemia.[7,8] Here, the phenolic metabolites
were determined from the collected urine and plasma samples
to clarify the molecular structures potentially affecting the
observed health effects. The acylated anthocyanins were served
as an extract free from potato matrix instead of comparing two
potato varieties in order to remove the effect of compositional
differences in phenolic compounds, starch, and monosaccha-
rides on the postprandial state and metabolites. According to our
knowledge, this is the first study to screen, identify, and quantify
the phenolic metabolites of acylated anthocyanins from serially
collected urine and plasma samples of humans.
2. Experimental Section
2.1. Solvents
For solid-phase extraction, formic acid (reagent grade, ≥95.0%)
andmethanol (HPLC grade,≥99.8%) were obtained from Sigma-
Aldrich (St Louis, MO). For LC-MS analyses, LC-MS grade ace-
tonitrile (Thermo Scientific, Rockford, IL, for anthocyanins and
CHROMASOLV, Sigma Aldrich, St Louis, MO, for other pheno-
lics), MS-grade formic acid (98%, Honeywell, Muskegon, MI),
and UPLC/MS-grade glacial acetic acid (Biosolve B.V., Dieuze,
France) were purchased. Ammonium acetate for HPLC buffer
preparation (HPLC-grade, ≥99.0%) was obtained from Sigma-
Aldrich (St Louis, MO). Water used was Milli-Q grade (Millipore,
Burlington, MA).
2.2. Standard Compounds
Anthocyanin standards (cyanidin-3-O-glucoside, cyanidin-
3-O-rutinoside, delphinidin-3-O-glucoside, delphinidin-3-O-
rutinoside, pelargonidin-3-O-glucoside, pelargonidin-3-O-
rutinoside, and malvidin-3-O-glucoside) were purchased from
Extrasynthese (Genay, France). Caffeine-(trimethyl-d9) (99
atom-% D, later caffeine-d9), catechol, 4-hydroxybenzaldehyde,
3-O-methyl-gallate, methyl vanillate, phloroglucinol, phloroglu-
cinaldehyde, pyrogallol, quercetin, quercetin-3-O-rutinoside,
protocatechuic aldehyde, taxifolin, as well as caffeic, chlorogenic,
ferulic, gallic, hippuric, homoprotocatechuic, homovanil-
lic, 3-hydroxybenzoic, 4-hydroxybenzoic, 5-hydroxyferulic,
4-hydroxyphenylacetic, isovanillic, 2-methoxybenzoic, 3-methyl
hippuric, 4-methyl hippuric, p-coumaric, protocatechuic, sinapic,
syringic, trans-cinnamic, and vanillic acids from Sigma-Aldrich
(St Louis, MO). Methyl-3,4-dihydroxybenzoate was purchased
from Alfa Aesar, Haverhill, MA. Gallic acid-3-O-glucuronide,
gallic acid-4-O-glucuronide, isoferulic acid-3-O-glucuronide,
protocatechuic acid-3-O-glucuronide, protocatechuic acid-4-O-
glucuronide, protocatechuic acid-3-O-sulfate, and protocatechuic
acid-4-O-sulfate were kindly synthesized by Dr Paul Needs at
The Quadram Institute Biosciences, Norwich, UK.
2.3. Ethics
The study protocol was accepted by the Ethical Committee of
theHospital District of Southwest Finland (ETMK:93/1801:2016)
and registered at clinicaltrials.gov as NCT02940080. Each study
subject provided written informed consent.
2.4. Study Design
Seventeen healthy men aged between 18 and 45 years consumed
350 g of steam-cooked yellow potato mash with (study meal)
and without (control meal) PPE solution rich in acylated antho-
cyanins as a breakfast after a 12-h fast in a cross-over, single-
blinded, and randomized study described previously. The PPE
and yellow potato mash were carefully characterized for nutri-
ent composition, sugars, organic acids, anthocyanins, and other
phenolic compounds.[8] For 48 h before and 24 h after consum-
ing a test meal, the participants followed a study diet composed
of foods and drinks low in flavonoids and dietary fiber. Total voids
of urine were collected at fasting state, then pooled at 0–4, 4–8, 8–
12, and 12–24 h postprandially. Urine samples (154 in total) were
stored as aliquots at −80 °C containing 1/0.2 V/V trifluoroacetic
acid (0.44 M). Venous blood samples (268 in total) were collected
into lithium-heparin tubes at fasting state, and then 20, 40, 60,
90, 120, 180, and 240 min postprandially. Blood samples were
centrifuged for 15 min, 1500 × g, to collect plasma, and then
acidified with 1/0.2 V/V trifluoroacetic acid (0.44 M) and stored
at −80 °C until analyses.
2.5. Dosage Information
The PPE was extracted from purple potatoes (Solanum tuberosum
L. ′Synkeä Sakari′) as described previously using aqueous ethanol
acidified with acetic acid and purified with Amberlite XAD-7HP
adsorbent.[27] The PPE was used instead of potatoes as such to
remove the effect of the compositional differences of potato va-
rieties. The PPE was carefully characterized earlier,[8] providing
152mg of anthocyanins and 140mghydroxycinnamic acid conju-
gates per meal corresponding to 0.48 kg of fresh purple potatoes.
The methoxysubstituted petunidin and peonidin derivatives
dominated with minor amounts of cyanidin, malvidin, delphini-
din, and pelargonidin. The anthocyanidins were linked to a glu-
cose and a rutinose (a disaccharide of a rhamnose and a glucose),
and 99% of the anthocyanins were monoacylated with a caffeic,
ferulic, or coumaric acid. Petunidin-coumaroyl-rutinoside-
glucoside represented 60% of the anthocyanins (Figure 1). The
hydroxycinnamic acid derivatives in the PPE were chlorogenic
acid (84%), caffeic acid (6%), and cryptochlorogenic acid (3%).
Bothmeals contained 0.7mg of flavonol glycosides and 4.5mg of
hydroxycinnamic acids from the yellow potatoes.[8] The amount
of anthocyanins was based on the previous clinical trial.[7]
Mol. Nutr. Food Res. 2021, 65, 2000898 2000898 (2 of 11) © 2021 The Authors. Molecular Nutrition & Food Research published by Wiley-VCH GmbH
www.advancedsciencenews.com www.mnf-journal.com
Petunidin-3-p-coumaroyl-
rutinoside-5-glucoside
Ferulic acid
Caffeic acid
p-Coumaric acid
Figure 1. The major anthocyanin of purple potatoes, petunidin-
coumaroyl-rutinoside-glucoside, and the acyl groups of purple potato
anthocyanins: caffeic, p-coumaric, and ferulic acids.[8] The suggested
positions of the acyl and glycosyl groups and the isomer of the coumaric
acid are based on a previous nuclear magnetic resonance spectroscopy
(NMR) characterization study.[48]
2.6. Sample Randomization
The samples of each volunteer were processed and analyzed as a
set of 10 (urine) and 16 (plasma) samples. Within these sample
sets, the samples were paired according to the sampling time
point with randomized order of the study visit. The order of
these pairs was randomized within the set, and the order of the
sample sets of the volunteers were randomized. The randperm
function in Matlab (The Mathworks, Cambridge, UK) was used
with the default setting of Mersenne Twister.
2.7. Solid-phase Extraction
The modified extraction method[28] with the chosen StrataX
(6 mL, 500 mg) solid-phase cartridges was verified by comparing
the analytical responses of matrix-matched blank samples spiked
with selected compounds before and after solid-phase extraction
(extraction recoveries, Supporting Information S1 and S2).
Samples were extracted in a room with yellow light to protect the
light-sensitive anthocyanins. The SPE cartridges were activated
and stabilized using one load volume of 0.25% formic acid in
methanol followed by a load volume of 0.1% formic acid in
water. Thawed samples were diluted 2/1 with 0.1% formic acid
and spiked with 10 µL of 400 mM a mid-polar and mid-range
responder, taxifolin, as an internal SPE process control standard.
The cartridges were washed twice with a load volume of 2%
methanol in 0.1% formic acid for urine, and 0.1% formic acid
for plasma. The cartridges were dried under vacuum for 30 min.
Samples were eluted with 5 mL 0.25% formic acid in methanol
after 10 min of soaking, followed by an additional 2 mL rinse.
The samples were evaporated to a volume of about 200 µL and
reconstituted with 200 µL of 0.1% formic acid. As an internal
volume control standard, 10 µL of caffeine-d9 (200 mM) was
added. The samples were shaken after which urine samples were
spun for 10 s and plasma samples centrifuged 10 000 × g for 10
min at 4 °C. For anthocyanin analyses, 200 µL of the supernatant
was acidified with 10 µL of strong formic acid. All samples were
stored in −80 °C. Prior to analyses, the samples were thawed,
shaken carefully, and centrifuged for 10 000 × g for 5 min at 4 °C.
Matrix-matched positive (urine) and standard-spiked (urine,
plasma) quality control samples were processed daily along with
the analytical samples.
2.8. UHPLC-MS/MS Analyses
Phenolic metabolites and anthocyanins were analyzed using
1290 Infinity UHPLC combined with 6490 Triple Quad mass
spectrometer equipped with iFunnel (Agilent Technologies,
Santa Clara, CA). The analytes were separated in an ACQUITY
UPLC HSS T3 1.8 µm (2.1 × 100 mm, Waters, Milford, MA) col-
umn at 35 °C. The injection volume was 5 µL for urine and 1 µL
for plasma. The sample rackwas cooled to 4 °C. Identification and
quantification of the phenolicmetabolites and anthocyaninswere
conducted with comprehensive, targeted multiple reaction mon-
itoring (MRM) panels (Supporting Information S1, S2), which
contained the degradants, metabolites and conjugates predicted
by the research group and presented in the current literature
for nonacylated anthocyanins. Authentic standard compounds
(Chapter 2.2) were used in identification whenever available.
For anthocyanins, the gradient consisted of 5% formic acid in
water (A) and 5% formic acid in acetonitrile (B) at 0.4 mL min−1
followingly: 0‒1 min, 5% B; 5 min, 10% B; 30 min, 25% B;
31 min, 95% B; 32 min, 95% B; 32.1 min, 5; % B and 36 min, 5%
B. Collision gas temperature was 220 °C, gas flow 14 L min−1,
nebulizer pressure 25 psi, capillary voltage 3.5 kV and collision
energy 5 eV for phosphatidyl choline and anthocyanidins, and
28 eV for anthocyanins. The retention times for the purple
potato anthocyanins were verified using urine spiked with the
PPE. For phenolic metabolites, the mobile phase A was 10 mM
ammonium acetate in water (pH 5) and B was 10 mM ammo-
nium acetate in acetonitrile (pH 5). The flow rate was 0.4 mL
min−1. The concentration of eluent B was 1% for a minute,
then at 3 min, 5%; 8 min, 60%; 8.50 min, 99% and 9–12 min,
1%. MRM was used in negative mode except for caffeine-d9 and
phosphatidyl choline. Collision gas temperature was 220 °C, gas
flow 14 L h−1, nebulizer 25 psi, and capillary voltage 3500 V.
Collision energies were optimized for each compound available.
The performance of the mass spectrometer was monitored by
analyzing repeatedly a matrix-matched sample containing all
available standard compounds.
The data was processed using MassHunter Quantitative Anal-
ysis B.06.00 (Agilent Technologies, Santa Clara, CA, USA). The
sample volumes were corrected on the basis of the caffeine-d9 lev-
els in each sample. Then, the collected data was divided into two
parts: the quantified metabolite data set, of which identification
of the compounds was confirmed using the available standard
compounds, and the screenedmetabolite data set, which was ten-
tatively identified based on the MRM transitions and chromato-
graphic retention. Quantification was performed using matrix-
matched external standard curves, and endogenously present
target metabolites were subtracted. All quantified metabolites
were within the linear range of the standard curves. If a metabo-
lite was detected in the samples of over half of the study volun-
teers (nine or more), it was interpreted as detected, and further
quantified.
Mol. Nutr. Food Res. 2021, 65, 2000898 2000898 (3 of 11) © 2021 The Authors. Molecular Nutrition & Food Research published by Wiley-VCH GmbH
www.advancedsciencenews.com www.mnf-journal.com
2.9. Urinary Creatinine
Urinary creatinine was analyzed in the laboratory of the Hospital
District of Southwest Finland using enzymatic methods (Cobas
C702 automatic analyzer, Roche Diagnostics GmbH,Mannheim,
Germany).
2.10. Pharmacokinetic and Statistical Analyses
For the separate time points of the quantified data set, statis-
tical analyses were performed using R 3.5.1.[29] For normally
distributed data (Shapiro–Wilk test), the paired-samples t-test
was used to compare the differences in the metabolite concen-
trations between the plasma and urine samples at certain time
points after the control and the study meal. The nonparametric
counterpart, the Wilcoxon signed rank test, was used otherwise.
The significance level was set to 0.05. Additionally, the pharma-
cokinetic incremental area under the curve (iAUC) values were
calculated using the trapezoidal rule, and maximum concen-
tration (Cmax) and time point at the maximum concentration
(tmax) were determined as the mean of the maximal values of
the individual volunteers. For urine, tmax was determined on the
basis of the most frequent categorical time period variable.
The screened and tentatively identified data set was visualized
as heatmaps using R 3.5.1[29] with gplots version 3.0.1.1[30] and
RColorBrewer version 1.1.2.[31] The differences between the vol-
ume corrected metabolite areas between the study and control
meals were calculated resulting in a variable describing change
between the meals in certain time point or period (delta). Due
to the large differences in the areas of the screened metabolites,
the fasting state value of the delta variable was set to zero by sub-
tracting it from the postprandial values. All compounds were nor-
malized to range [−1, 1] by dividing with the absolute maximum
delta of the corresponding compound. The time-wise change of
the delta variable of the compounds regarding to the fasting state
can be observed, but the change is not comparable between the
compounds.
3. Results and Discussion
3.1. Anthocyanins
In urine, five anthocyanin degradants were detected: cyanidin-3-
O-glucoside, as confirmed with the standard compound, and
tentatively identified, malvidin-rutinoside, petunidin-glucoside,
peonidin-glucoside, and peonidin-glucuronide (aglycones, Fig-
ure 2; heatmap Figure 3A). The anthocyanin degradants peaked
at 8–12 h, and were present in urine samples until 24 h, suggest-
ing enterohepatic circulation of the degradants. These results in-
dicate that the monoacyl group and rutinose are hydrolyzed. De-
tection of cyanidin-3-O-glucoside refers to the high bioavailability
of theminor cyanidin-based anthocyanins present in potato orO-
demethylation of the methoxysubstituted peonidin to cyanidin.
Detecting peonidin-glucuronide, a conjugate, indicates phase II
metabolism. A subset of plasma samples was analyzed for antho-
cyanins, but none were detected.
This is the first clinical study to detect degradants of monoacy-
lated anthocyanins of purple potatoes in human urine;[26] frag-
ments of cyanidin-derived mono- and diacylated anthocyanins
of red cabbage have been detected in earlier clinical trials.[5,32]
The nonacylated monoglycosylated anthocyanins are deglycosy-
lated after ingestion and extensively glucuronidated and methy-
lated, and also sulfated in minor amounts,[22,28] leading to 91%
of the total anthocyanins appearing as conjugates of aglycones
in urine.[22] On the contrary, our results show that the absorbed
potato acylated anthocyanins seem to be resistant for deglycosy-
lation. Nonacylated anthocyanin conjugates have been detected
after a meal of red cabbage rich in acylated anthocyanins,[5,32] but
according to our knowledge, this is the first time glucuronida-
tion is reported for acylated anthocyanins derived from purple
potatoes.[26]
Acylation decreases the bioavailability of anthocyanins[5,6]
possibly due to the steric hindrance of several glycosyl and
acyl moieties against the gastric absorption mechanisms as
detected in vitro.[33,34] Selective structure-dependent absorption
of anthocyanins has been suggested,[12] but understanding the
dependence requiresmore investigations as contradictory results
have been presented. Previous clinical interventions which fed
purple potatoes[26] and Concord grape juice[24] (38% of total an-
thocyanins were monoacylated, mainly delphinidin, petunidin,
and malvidin diglucosides) to healthy volunteers either did not
detect acylated anthocyanins[24,26] or any anthocyanins at all[26] in
human biofluids. Other clinical interventions have detected se-
lected, structurally different (B-ring, acylation pattern, sugarmoi-
ety) acylated anthocyanins of red cabbage (mono- and diacylated
cyanidin diglucosides),[5,32] purple carrot (monoacylated cyanidin
xylosides),[6] and purple sweet potatoes (diacylated cyanidin and
peonidin sophorosides)[12,13] in human plasma and/or urine, and
in addition, of eggplant (monoacylated delphinidin rutinosides)
in rats.[35] As our methodology was optimized with nonacylated
anthocyanins and the level of acylated anthocyanins in biological
fluids is low due to their extremely poor bioavailability, more
studies are needed to confirm our results.
The B-ring substitution and sugar moieties of the ingested
anthocyanins affect the bioavailability in vivo.[4,22,36] In humans,
increased hydroxylation of B-ring (delphinidin, cyanidin) may
increase the bioavailability of nonacylated anthocyanins[36] but it
should be noted that anthocyanidins are readily interconverted
to each other in phase I and II reactions via demethylation,
methylation, hydroxylation, and dihydroxylation reactions in
vivo. For example, the methylation of nonacylated cyanidin to
peonidin has been well-recognized,[14,22,35] whereas malvidin
may be demethylated to petunidin.[22] The anthocyanidin may
also affect the degree of enzymatic conjugation; for malvidin,
it may be decreased but for the slightly smaller petunidin, the
conjugation is similar compared to cyanidin glycosides.[22] In
future, studies feeding pure labeled anthocyanins are ideal for
further structural comparisons.
3.2. Quantified Phenolic Metabolites
From urine, 28 phenolic metabolites were identified and quan-
tified (Figure 2, Figure 4A, Table 1, Supporting Information S3,
S5). The mean maximum concentrations (Cmax) varied from
2 nM mM−1 creatinine of quercetin to 290 µM mM−1 creatinine
of hippuric acid, and the tmax values of the detected metabolites
were late (4–8, 8–12 h). Of the urinary metabolites, 19 were
Mol. Nutr. Food Res. 2021, 65, 2000898 2000898 (4 of 11) © 2021 The Authors. Molecular Nutrition & Food Research published by Wiley-VCH GmbH
www.advancedsciencenews.com www.mnf-journal.com
Enzymes
pH
Gut microbiota
Acylated anthocyanins from potatoes
R1 = OH & R2 = H, cyanidin
R1 = OH & R2 = OH, delphinidin
R1 = OCH3 & R2 = OCH3, malvidin
R1 = H & R2 = H, pelargonidin
R1 = OCH3 & R2 = H, peonidin
R1 = OCH3 & R2 = OH, petunidin
Hippuric acid (U, P)
Catechol (U)
B B
4-Hydroxybenzoic acid 
(U, P)
Phloroglucinol (U) Phloroglucinaldehyde (U)
A
Isoferulic acid-3-O-glucuronide (U, P)
Hydrolysis (R3, R4)
Ring C fission
Phase I, II reactions
4-Hydroxyphenylacetic 
acid (U)
B
Protocatechuic acid conjugates (U)
R1 = OH & R2 = OH, PCA
R1 = OGlc & R2 = OH, PCA-3-O-Glc
R1 = OH & R2 = OGlc, PCA-4-O-Glc
R1 = OSO3- & R2 = OH, PCA-3-O-S
R1 = OH & R2 = OSO3-, PCA-4-O-S
B
Hydroxycinnamic acids
R1= H & R2 = H, p-coumaric acid (U)
R1= H & R2 = OH, caffeic acid (U)
R1= H & R2 = OCH3, ferulic acid (U)
R1= OCH3 & R2 = OCH3, sinapic acid (U)
B B
B
A C
B
Gallic acid (U)
B
Vanillic acid (U, P)
A
Figure 2. Identified phenolic metabolites detected in urine (U) and plasma (P) after a meal of purple potato extract rich in acylated anthocyanins and
their suggested origin. PCA, protocatechuic acid; glc, glucuronide; s, sulfate. R3 and R4 may carry a glycoside of which R3 may be acylated with a
hydroxycinnamic acid. The values are presented in the Supporting Information S3 and S4.
elevated statistically significantly after the study meal compared
to the control meal. Hippuric acid reached an increase of 139 µM
mM−1 creatinine from the control levels after the study meal
(p = 0.026 and 0.006 at 4–8 and 8–12 h). Monomethoxysub-
stituted vanillic acid (p = 0.002 and 0.001 at 0–4 and 4–8 h,
respectively) and dimethoxysubstituted sinapic acid (p = 0.000
at 4–8 h) were detected. Other elevated abundant metabolites
were hydroxycinnamic acids, including caffeic acid (p = 0.006
and 0.003 at 4–8 and 8–12 h), p-coumaric acid (p = 0.014 at
4–8 h), and ferulic acid (p = 0.041, <0.001, and 0.010 at 0–4,
4–8, and 8–12 h). Dihydroxysubstituted protocatechuic acid was
increased (p = 0.003, 0.003, and 0.001 at 0–4, 4–8, and 8–12 h,
respectively). Furthermore, its 3- and 4-sulfates and glucuronides
were detected, of which protocatechuic acid-4-O-glucuronide,
-3-O-sulfate and -4-O-sulfate were increased (p = 0.008, 0.017,
0.025 at 0–4 , 4–8, and 8–12 h; p = 0.020, 0.001, and 0.010 at 0–4,
Mol. Nutr. Food Res. 2021, 65, 2000898 2000898 (5 of 11) © 2021 The Authors. Molecular Nutrition & Food Research published by Wiley-VCH GmbH
www.advancedsciencenews.com www.mnf-journal.com
A B
C
Figure 3. Screened A) anthocyanins in urine, B) phenolic metabolites in urine, and C) phenolic metabolites in plasma of 17 healthy men after a potato
meal with (study meal) or without (control meal) anthocyanin-rich purple potato extract. Tentative identifications are based on the MRM transition
and chromatographic retention. The heatmaps describe change between the samples collected after the study and control meal at certain time points
standardized to the range of [−1, 1]. Purple color indicates that larger amount of the metabolites were detected after the study meal, whereas orange
indicates the opposite.
Mol. Nutr. Food Res. 2021, 65, 2000898 2000898 (6 of 11) © 2021 The Authors. Molecular Nutrition & Food Research published by Wiley-VCH GmbH
www.advancedsciencenews.com www.mnf-journal.com
Figure 4. The phenolic metabolites identified and quantified in A) urine and B) plasma of 17 healthy men using corresponding standard compounds.
Values are given as concentrations (nMmM−1 creatinine, urine and nM, plasma) with standard deviations in the fasting state (0-point) and postprandially
(Figure A, h; Figure B, min). The purple plots represent the treatment values whereas the orange plots represent the control values. The between-meal
significant differences (p < 0.05) are shown with asterisks. Supporting information S5 and S6 contain the p-values.
Mol. Nutr. Food Res. 2021, 65, 2000898 2000898 (7 of 11) © 2021 The Authors. Molecular Nutrition & Food Research published by Wiley-VCH GmbH
www.advancedsciencenews.com www.mnf-journal.com
Table 1. The urinary pharmacokinetic parameters of the phenolic metabolites after an acute intake of a purple potato extract rich in methoxysubstituted,
monoacylated anthocyanins.
n Cmax [nM mM
−1 creatinine] SD tmax [h] Detected earlier in
Caffeic acid 17 93.6 71.0 4–8 [16,17,19,21]
Catechol 13 41.7 17.4 4–8
Chlorogenic acid
a
17 131.8 99.9 4–8 [19]
p-Coumaric acid 14 85.0 73.9 8–12 [16–19,21,24,28]
Ferulic acid 17 119.9 67.5 4–8 [3,14–18,21,23,28]
Gallic acid 14 1.8 0.8 4–8 [17,21]
Gallic acid-4-O-glucuronide 17 68.6 40.4 0–4
Hippuric acid 17 288 692.8 166 291.8 8–12 [3,15,17,19,21,23,26,28]
Homoprotocatechuic acidb 14 26.6 19.6 4–8 [3,14,16–19,21,23,28]
Homovanillic acidc 17 538.1 437.5 4–8 [16–19,21,23,25,28]
4-Hydroxybenzaldehyde 17 200.0 212.6 8–12 [3,17,19,21]
3- & 4-Hydroxybenzoic acid 17 1791.7 840.4 4–8 [3,16–19,21,23,28]
4-Hydoxyphenylacetic acidd 17 266.3 170.4 0–4 [3,14,17,19,21]
Isoferulic acid-3-O-glucuronide 17 345.6 255.8 4–8 [17,19,23,24]
Isovanillic acid 15 140.2 117.5 0–4 [3,14,17,19,21]
2-Methoxybenzoic acide 16 34.7 73.5 8–12
Methyl-3,4-dihydroxybenzoate 12 1.7 1.9 4–8 [3,14,21]
3- & 4-Methylhippuric acid 14 560.4 1273.3 8–12 [28]
Phloroglucinaldehyde 16 3.5 1.9 4–8 [3,18,28]
Phloroglucinol 15 86.4 49.9 0–4 [21]
Protocatechualdehyde 17 5.9 5.0 8–12 [3,19,21]
Protocatechuic acid 17 53.8 32.3 4–8 [3,14–19,21,23]
Protocatechuic acid-4-O-glucuronide 15 8.8 9.1 4–8 [3,14,18,28]
Protocatechuic acid-3-O-sulfate 17 482.3 250.6 4–8 [3,14,18,21,28]
Protocatechuic acid-4-O-sulfate 17 239.4 418.3 0–4 [3,14,18,21,28]
Quercetin
a
17 1.5 1.9 4–8 [17,19]
Sinapic acid 17 53.7 36.1 4–8 [17–19,28]
Vanillic acid 17 676.7 412.9 0–4 [14,16–19,21,23,25,28]
The values are presented as mean with standard deviations (SD) detected in n study participants with the exception of the categorical tmax, which is presented on the basis
of the highest frequency. Cmax, the mean of the maximum metabolite concentrations of the volunteers; tmax was determined on the basis of the highest frequency of the
categorical time period variable;
a
Chlorogenic acid and quercetin are not likely anthocyanin metabolites but are derived from the meals; Quantified with bprotocatechuic acid;
cvanillic acid; dcaffeic acid; egallic acid.
4–8, and 8–12 h; p = 0.003, 0.002, and <0.001 at 0–4, 4–8, and
8–12 h, respectively). Protocatechuic acid-3-O-glucuronide was
detected but not quantified due to co-elution.
Of the 14 phenolic metabolites detected and quantified from
plasma (Figure 2, Figure 4B, Table 2, Supporting Information
S4 and S6), five were statistically significantly elevated after the
study meal compared to the control meal. Chlorogenic acid was
increased at 60, 120, and 180 min postprandially (p = 0.024,
<0.001, and 0.022, respectively), hippuric acid at 0, 120, and
240min (p= 0.039, 0.001, 0.038), isoferulic acid-3-O-glucuronide
at 120 and 180 min (p = 0.018 and 0.013), quercetin at 90 min (p
= 0.038), and vanillic acid at 90 min (p = 0.036). The Cmax ranged
between 2 nM of methyl-3,4-dihydroxybenzoate and 4700 nM of
hippuric acid, whereas most of the tmax values were between 1
and 2 h. Plasma contained fewer metabolites compared to urine,
to which the kidneys concentrate metabolites for excretion. Late-
rise metabolites, however, were not detected from plasma due to
the 4-h postprandial sampling.
These results show that the acylated anthocyanins of purple
potatoes are degraded into phenolic metabolites in humans. O-
Demethylation of the metabolites may occur as the potato an-
thocyanins are mainly methoxysubstituted, but dihydroxylated
catechol and protocatechuic acid and its conjugates were abun-
dant in urine. Detecting conjugated phenolic acids indicates
phase II metabolism as suggested in studies regarding berry
anthocyanins.[17–19,28] Hippuric acid, the major metabolite of
potato anthocyanins, is another phase II metabolite formed by
the liver cells from benzoic acid and glycine, and is a recog-
nized biomarker for consumption of polyphenol-rich berries
and fruits.[37,38] However, the levels of hippuric acid are affected
also by endogenous, such as protein, metabolism. Both urine
and plasma contained p-coumaric acid, and in urine, also fer-
ulic acid and caffeic acid were abundant; they might be orig-
inated from the A- and B-rings and the acyl groups of the
parent anthocyanins. p-Coumaric acid and caffeic acid were
both also present in minor amounts in the ingested PPE, and
Mol. Nutr. Food Res. 2021, 65, 2000898 2000898 (8 of 11) © 2021 The Authors. Molecular Nutrition & Food Research published by Wiley-VCH GmbH
www.advancedsciencenews.com www.mnf-journal.com
Table 2. The plasma pharmacokinetic parameters of the phenolic metabolites after an acute intake of a purple potato extract rich in methoxysubstituted,
monoacylated anthocyanins.
n Cmax [nM] SD tmax [h] SD iAUC0-4 [nM*h L
−1] SD Detected earlier in
Caffeic acid 17 27.1 18.9 1.7 1.2 14.4 12.3 [17,19,24]
Chlorogenic acid
a
16 36.4 24.0 2.5 0.8 59.9 53.9 [17,19]
Hippuric acid 17 4649.2 1293.3 1.7 1.6 370.6 611.6 [3,15,17,19,21,28]
4-Hydroxybenzaldehyde 17 264.2 122.7 1.2 1.0 40.9 42.8 [3,17–19,21,28]
3- & 4-Hydroxybenzoic acid 17 214.3 123.7 1.4 1.3 163.7 175.8 [17,19,21,28]
4-Hydroxyphenylacetic acid
b
12 23.1 17.0 1.3 1.2 12.8 13.0 [17,19,21]
Isoferulic acid-3-O-glucuronide 15 161.2 130.2 2.2 1.2 240.6 173.8 [17,19,23]
Methyl-3,4-dihydroxybenzoate 17 1.8 1.1 1.7 1.4 1.6 1.7 [3,14,18,21,28]
Phloroglucinaldehyde 12 13.3 7.0 1.6 1.3 14.5 17.7 [4,18,21,28]
Protocatechualdehyde 17 12.3 6.5 1.1 1.1 9.0 9.6 [18,19,21,28]
Protocatechuic acid 16 7.9 2.9 1.8 1.3 7.0 6.9 [3,4,14,15,17–19,21,28]
Protocatechuic acid-3-O-sulfate 16 15.4 8.2 1.6 0.9 13.6 15.8 [3,14,18,28]
Quercetin
a
15 6.4 4.0 1.8 1.1 7.3 7.5 [17]
Vanillic acid 12 91.5 38.4 1.4 0.7 76.0 71.6 [3,4,14,18,19,28]
The values are presented asmean with standard deviations (SD) detected in n study participants. Cmax, the mean of the maximummetabolite concentrations of the volunteers;
tmax, mean of the time point when Cmax was observed; iAUC, incremental area under the curve calculated using the trapezoidal rule;
a
Chlorogenic acid and quercetin are not
likely anthocyanin metabolites but are derived from the meals;
b
Quantified with caffeic acid.
chlorogenic acid, abundant in the PPE, may partially break down
to caffeic acid. Ferulic acid may be formed from caffeic acid via
O-methylation.[39] Chlorogenic acid and quercetin aremost likely
derived from the PPE and yellow potatoes[8] and are not metabo-
lites of anthocyanins.
Under physiological conditions, nonacylated anthocyanins
are rapidly hydrolyzed leading to the formation of an unstable 𝛼-
diketone which, without the protecting glycoside, is subjected to
degradation leading to a phenolic acid (expected B-ring metabo-
lite) and aldehyde (expected A-ring metabolite).[40] Thereafter,
the phenolic metabolites may be conjugated as detected with la-
beled cyanidin-3-O-glucoside in vivo.[3,14] Acylated anthocyanins,
on the other hand, are structurally more stable than nonacylated
anthocyanins as the acyl groups protect from the nucleophilic
attack of water, and in comparison to hydroxysubstituted an-
thocyanins, methoxysubstituted are less reactive. Even 72% of
the applied acylated anthocyanins of potatoes (but only 45%
of the cyanidin-based, acylated anthocyanins of carrot) may
reach the colonic vessel in an in vitro gastrointestinal model.[41]
In colon, acylated anthocyanins may be degraded by human
gut microbiota as detected in vitro with mono- and diacylated
pelargonidin sophoroside glucosides.[40] This is substantiated
by our in vivo results: in addition to late tmax times, urinary pro-
tocatechuic acid and phloroglucinaldehyde, the known colonic
metabolites of cyanidin-3-O-glucoside,[42] were detected. The
proposed degradation mechanism occurs first by enzymatic
hydrolysis of the acyl and glycosyl groups followed by enzymatic
and/or spontaneous C-ring fission of the unstable anthocyanidin
leading to phenolic metabolites, which may be then subjected to
further phase I and II reactions (Figure 2).
Considering the earlier reported beneficial health effects
of purple potatoes[7] and the PPE[8] on human carbohy-
drate metabolism, acylated anthocyanins may inhibit pancre-
atic 𝛼-amylase[43] and intestinal maltase[44] as detected in vitro.
More bioavailable phenolic metabolites may contribute to these
health effects; for example, protocatechuic acid has insulin-like
properties[45,46] and other phenolic metabolites may modulate
carbohydrate metabolism by inhibiting salivary 𝛼-amylase, en-
hancing the intestinal uptake of glucose and downregulating
sugar transporters as detected in vitro.[47] More investigations of
the physiological role of the phenolic metabolites are required to
understand the health effects and underlying molecular mecha-
nisms of anthocyanin-rich foods.
Metabolites not previously reported in clinical trials feeding
anthocyanin-rich foods were detected: catechol, gallic acid-
4-O-glucuronide, and 2-methoxybenzoic acid, of which only
catechol was increased statistically significantly in urine after
the study meal. The lack of statistical significance may be due
to inter-individual variation amongst the volunteers or that the
metabolite was derived from the yellow potatoes used both in
the control and study meals. Furthermore, phloroglucinol was
detected earlier by only one study,[21] which fed raspberries rich
in both anthocyanins and other phenolic compounds to healthy
human volunteers. Earlier, a meal of purple potatoes led to the
tentative identification of homovanillic acid/dihydrocaffeic acid,
dihydroferulic acid, (iso)ferulic acid, dihydrocaffeic acid sulfate,
and hippuric acid, but the study lacked the control arm.[26]
As a next-step suggestion, a trial feeding labelled acylated
anthocyanins would further discriminate the potato-derived
metabolites from anthocyanin-derived metabolites. Here,
large inter-individual variation was detected;[15,17–19] individual
metabolism and gut microbiota composition may affect the
metabolism of anthocyanins.
3.3. Screening of Tentatively Identified Phenolic Metabolites
A large number of phenolic metabolites and conjugates were
detected, tentatively identified and heatmapped (Figure 3B),
most of them peaking at 8–12 h. For example, vanillaldehyde
Mol. Nutr. Food Res. 2021, 65, 2000898 2000898 (9 of 11) © 2021 The Authors. Molecular Nutrition & Food Research published by Wiley-VCH GmbH
www.advancedsciencenews.com www.mnf-journal.com
and hydroxyhippuric acids, and isomers of hydroxybenzoic acid,
dihydroferulic acid, and dihydrosinapic acid were detected. Sul-
fates and glucuronides were detected for hydroxycinnamic acid
derivatives (coumaric, caffeic, and ferulic acids), benzoic acid
derivatives (hydroxybenzoic, vanillic, syringic, and methylgallic
acids), phenol derivatives (catechol and pyrogallol), phenylacetic
acid derivatives (homoprotocatechuic acid), and phenylpropanoic
acids such as dihydro(iso)ferulic and dihydrosinapic acids. From
plasma, 16 tentatively identified metabolites were detected
(Figure 3C). 2-Hydroxybenzoic acid showed high analytical
response compared to the other ones, peaking at 90 min after
the study meal. Several sulfate and glucuronide conjugates were
detected: catechol glucuronide, catechol sulfate, homoprotocate-
chuic acid glucuronide, hydroxybenzoic acid sulfate, (iso)ferulic
acid glucuronide isomer, two methoxycatechol sulfate isomers,
methylgallic acid sulfate, and vanillic acid sulfate.
The screened data set highlights the vast extent of phenolic
metabolites after an acute intake of PPE rich in acylated an-
thocyanins even though the identification was only tentative
due to practical challenges. The number of possible degradants
and conjugates to be screened was large as the potato antho-
cyanins contain all six anthocyanidins varying in the acylation
pattern, but the availability of commercial standards is limited.
Therefore, in addition to our large compound library, the MRM
transitions and chromatographic retention were used for identi-
fying the screened dataset as reported earlier.[22] Additionally, the
concentration of anthocyanins and phenolic metabolites is low
in physiological samples. Here their levels were concentrated
using optimized SPE and evaporation, and the samples were
analyzed using a modern state-of-the-art mass spectrometer
with a targeted MRM approach providing high instrumental
sensitivity and capacity to scan numerous transitions without
compromising the amount of data points per peak. Regardless
of its high sensitivity, a targeted MRM approach is limited to
predicted metabolites, whereas an untargeted approach, which
is usually performed with high resolution, could lead to the
identification of unpredicted metabolites.
4. Concluding Remarks
This report demonstrates that an acute intake of PPE rich in
methoxysubstituted anthocyanins monoacylated with caffeic,
coumaric and ferulic acids leads to absorption and excretion
of only a small number of anthocyanins degradants and an
extensive number of spontaneous and/or colonic phenolic
metabolites and their phase II conjugates. The rutinose and acyl
groups of anthocyanins are hydrolyzed and the methoxysubsti-
tuted anthocyanidins may be O-demethylated. Novel phenolic
metabolites of acylated potato anthocyanins, such as glucuronyl
and sulfonyl conjugates of protocatechuic acid, were detected.
Three novel phenolic metabolites not reported earlier in clinical
trials after a meal rich in anthocyanins, such as purple pota-
toes, were detected: catechol, gallic acid-4-O-glucuronide, and
2-methoxybenzoic acid, of which only catechol was increased
statistically significantly in comparison to the control yellow
potato meal without the PPE. These results provide new insights
on the bioavailability and metabolism of acylated anthocyanins
in humans.
Supporting Information
Supporting Information is available from the Wiley Online Library or from
the author.
Acknowledgements
This study was funded by the Doctoral Programme of Molecular Life
Sciences of the University of Turku Graduate School, Finland, the Orion
Research Foundation sr, Finland, ERVA of the City of Turku, Finland, the
Raisio Research Foundation, Finland, the Magnus Ehrnrooth Foundation,
Finland, the Biotechnology and Biological Sciences Research Council
(BBSRC, UK) through an Institute Strategic Programme Grant (“Food
Innovation and Health”; Grant No. BB/R012512/1 and its con-
stituent projects BBS/E/F/000PR10343, BBS/E/F/000PR10345 and
BBS/E/F/000PR10346) to the Quadram Institute Biosciences and a
Norwich Research Park Biosciences Doctoral Training Partnership award
to JP (grant number BB/M011216/1), and the Niemi foundation, Finland.
The volunteers are thanked for their participation. MD Matti Viitanen
is thanked for ethical and medical expertise, Sanna Himanen for blood
sampling (the City Hospital of Turku, Finland). Dr Paul Needs is thanked
for synthesis of the standard compounds, Dr Shikha Saha for analyt-
ical advice and Dr Marianne Defernez for sample randomization (the
Quadram Institute Bioscience, UK). Tuomas Jokioja is thanked for his
help with R. Dr Jari Heinonen and Professor Tuomo Sainio are thanked
for extracting the PPE (the Lappeenranta-Lahti University of Technology
LUT, Finland), and Dr Mikko Griinari for potato cultivation.
Author Contribution
J.J., K.L., and B.Y. concepted the study and organized the clinical trial. J.J.
and J.P. optimized the analysis method and collected the data under su-
pervision of M.P. and P.K., and J.J. processed the data. J.J., B.Y., P.K., and
K.L. participated to funding acquisition and B.Y., P.K., and K.L. provided
the resources. J.J. drafted the original manuscript with the help of K.L.,
and all authors commented on and approved the manuscript.
Conflict of interest
The authors declare no conflict of interest.
Data Availability Statement
Research data are not shared.
Keywords
acylated anthocyanins, mass spectrometry, postprandial metabolites,
purple-fleshed potatoes, Solanum tuberosum L.
Received: September 15, 2020
Revised: January 11, 2021
Published online: March 30, 2021
[1] K. Hanhineva, R. Törrönen, I. Bondia-Pons, J. Pekkinen, M.
Kolehmainen, H. Mykkänen, K. Poutanen, Int. J. Mol. Sci. 2010, 11,
1365.
[2] G. Williamson,Mol. Nutr. Food Res. 2013, 57, 48.
Mol. Nutr. Food Res. 2021, 65, 2000898 2000898 (10 of 11) © 2021 The Authors. Molecular Nutrition & Food Research published by Wiley-VCH GmbH
www.advancedsciencenews.com www.mnf-journal.com
[3] R. M. de Ferrars, C. Czank, Q. Zhang, N. P. Botting, P. A. Kroon, A.
Cassidy, C. D. Kay, Br. J. Pharmacol. 2014, 171, 3268.
[4] D. Mueller, K. Jung, M. Winter, D. Rogoll, R. Melcher, E. Richling,
Food Chem. 2017, 231, 275.
[5] C. S. Charron, B. A. Clevidence, S. J. Britz, J. A. Novotny, J. Agric. Food
Chem. 2007, 5354.
[6] A. C. Kurilich, B. A. Clevidence, S. J. Britz, P. W. Simon, J. A. Novotny,
J. Agric. Food Chem. 2005, 53, 6537.
[7] K.M. Linderborg, J. E. Salo,M. Kalpio, A. L. Vuorinen,M. Kortesniemi,
M. Griinari, M. Viitanen, B. Yang, H. Kallio, Int. J. Food Sci. Nutr. 2016,
67, 581.
[8] J. Jokioja, K. M. Linderborg, M. Kortesniemi, A. Nuora, J. Heinonen,
T. Sainio, M. Viitanen, H. Kallio, B. Yang, Food Chem. 2020, 125797.
[9] C. D. Kay, P. A. Kroon, A. Cassidy,Mol. Nutr. Food Res. 2009, 53, S92.
[10] K. Keppler, H.-U. Humpf, Bioorg. Med. Chem. 2005, 13, 5195.
[11] B. A. Cevallos-Casals, L. Cisneros-Zevallos, Food Chem. 2004, 86, 69.
[12] K. Harada, M. Kano, T. Takayanagi, O. Yamakawa, F. Ishikawa, Biosci.
Biotechnol. Biochem. 2004, 68, 1500.
[13] T. Oki, I. Suda, N. Terahara, M. Sato, M. Hatakeyama, Biosci. Biotech-
nol. Biochem. 2006, 70, 2540.
[14] C. Czank, A. Cassidy, Q. Zhang, D. J.Morrison, T. Preston, P. A. Kroon,
N. P. Botting, C. D. Kay, Am. J. Clin. Nutr. 2013, 97, 995.
[15] L. Xie, S. G. Lee, T. M. Vance, Y. Wang, B. Kim, J.-Y. Lee, O. K. Chun,
B. W. Bolling, Food Chem. 2016, 211, 860.
[16] T. Nurmi, J. Mursu, M. Heinonen, A. Nurmi, R. Hiltunen, S. Vouti-
lainen, J. Agric. Food Chem. 2009, 57, 2274.
[17] R. P. Feliciano, G. Istas, C. Heiss, A. Rodriguez-Mateos, Mol. Basel
Switz. 2016, 21, 1120.
[18] R. M. de Ferrars, A. Cassidy, P. Curtis, C. D. Kay,Mol. Nutr. Food Res.
2014, 58, 490.
[19] R. P. Feliciano, A. Boeres, L. Massacessi, G. Istas, M. R. Ventura, C.
Nunes dos Santos, C. Heiss, A. Rodriguez-Mateos, Arch. Biochem.
Biophys. 2016, 599, 31.
[20] I. Iswaldi, D. Arráez-Román, A. M. Gómez-Caravaca, M. del M. Con-
treras, J. Uberos, A. Segura-Carretero, A. Fernández-Gutiérrez, Food
Chem. Toxicol. 2013, 55, 484.
[21] X. Zhang, A. Sandhu, I. Edirisinghe, B. Burton-Freeman, Food Funct.
2018, 9, 806.
[22] W. Kalt, Y. Liu, J. E. McDonald, M. R. Vinqvist-Tymchuk, S. A. E. Fill-
more, J. Agric. Food Chem. 2014, 62, 3926.
[23] I. A. Ludwig, P. Mena, L. Calani, G. Borges, G. Pereira-Caro, L. Bres-
ciani, D. Del Rio,M. E. J. Lean, A. Crozier, Free Radical Biol. Med. 2015,
89, 758.
[24] A. Stalmach, C. A. Edwards, J. D. Wightman, A. Crozier, J. Agric. Food
Chem. 2011, 59, 9512.
[25] A. Stalmach, C. A. Edwards, J. D. Wightman, A. Crozier, Food Funct.
2013, 4, 52.
[26] C. Tsang, N. F. Smail, G. J. McDougall, S. Almoosawi, E. A. Al-Dujaili,
EC Nutr. 2015, 1.3, 96.
[27] J. Heinonen, H. Farahmandazad, A. Vuorinen, H. Kallio, B. Yang, T.
Sainio, Food Bioprod. Process. 2016, 99, 136.
[28] R. M. de Ferrars, C. Czank, S. Saha, P. W. Needs, Q. Zhang, K. S.
Raheem, N. P. Botting, P. A. Kroon, C. D. Kay, Anal. Chem. 2014, 86,
10052.
[29] R. Core Team, 2018. R: A language and environment for statistical
computing. R Foundation for Statistical Computing, Vienna, Austria.
https://www.R-project.org/
[30] G. R. Warnes, B. Bolker, L. Bonebakker, R. Gentleman, W. Huber,
A. Liaw, T. Lumley, M. Maechler, A. Magnusson, S. Moeller, M.
Schwartz, B. Venables, 2019, https://CRAN.R-project.org/package=
gplots
[31] E. Neuwirth, 2014. RColorBrewer: ColorBrewer Palettes. R package
version 1.1-2. https://CRAN.Rproject.org/package=RColorBrewer
[32] W. Wiczkowski, D. Szawara-Nowak, J. Romaszko, Food Chem. 2016,
190, 730.
[33] H. Oliveira, C. Roma-Rodrigues, A. Santos, B. Veigas, N. Brás, A.
Faria, C. Calhau, V. de Freitas, P. V. Baptista, N. Mateus, A. R. Fer-
nandes, I. Fernandes, Sci. Rep. 2019, 9, 789.
[34] S. Passamonti, U. Vrhovsek, F. Mattivi, Biochem. Biophys. Res. Com-
mun. 2002, 296, 631.
[35] T. Ichiyanagi, N. Terahara, M. M. Rahman, T. Konishi, J. Agric. Food
Chem. 2006, 54, 5306.
[36] T. K. McGhie, G. D. Ainge, L. E. Barnett, J. M. Cooney, D. J. Jensen, J.
Agric. Food Chem. 2003, 51, 4539.
[37] D. Krupp, N. Doberstein, L. Shi, T. Remer, J. Nutr. 2012, 142, 1314.
[38] J. Toromanovi´c, E. Kovacˇ-Bešovi´c, A. Šapcˇanin, I. Tahirovi´c, Z. Rim-
papa, G. Kroyer, E. Sofic´, Bosn. J. Basic Med. Sci. 2008, 8, 38.
[39] S. Lafay, C. Morand, C. Manach, C. Besson, A. Scalbert, Br. J. Nutr.
2006, 96, 39.
[40] J. Fleschhut, F. Kratzer, G. Rechkemmer, S. E. Kulling, Eur. J. Nutr.
2006, 45, 7.
[41] H. Zhang, Y. I. Hassan, J. Renaud, R. Liu, C. Yang, Y. Sun, R. Tsao,
Mol. Nutr. Food Res. 2017, 61, 1600928.
[42] K. Keppler, H.-U. Humpf, Bioorg. Med. Chem. 2005, 13, 5195.
[43] J. A. H. Kaeswurm, L. Könighofer, M. Hogg, A. Scharinger, M. Buch-
weitz, Foods 2020, 9, 367.
[44] T. Matsui, T. Ueda, T. Oki, K. Sugita, N. Terahara, K. Matsumoto, J.
Agric. Food Chem. 2001, 49, 1948.
[45] B. Scazzocchio, R. Varì, C. Filesi, I. Del Gaudio, M. D’Archivio, C. San-
tangelo, A. Iacovelli, F. Galvano, F. R. Pluchinotta, C. Giovannini, R.
Masella,Mol. Nutr. Food Res. 2015, 59, 1472.
[46] B. Scazzocchio, R. Varì, C. Filesi, M. D’Archivio, C. Santangelo, C. Gio-
vannini, A. Iacovelli, G. Silecchia, G. Li Volti, F. Galvano, R. Masella,
Diabetes 2011, 60, 2234.
[47] S. K. Barik, W. R. Russell, K. M. Moar, M. Cruickshank, L. Scobbie, G.
Duncan, N. Hoggard, J. Nutr. Biochem. 2020, 78, 108325.
[48] T. Fossen, D. O. Øvstedal, R. Slimestad, Ø. M. Andersen, Food Chem.
2003, 81, 433.
Mol. Nutr. Food Res. 2021, 65, 2000898 2000898 (11 of 11) © 2021 The Authors. Molecular Nutrition & Food Research published by Wiley-VCH GmbH