The fate of acylated anthocyanins in mildly heated neutral solution

Article abstract of DOI:10.1016/j.dyepig.2020.108326

In neutral solution, anthocyanins acylated by hydroxycinnamic acids typically exhibit attractive blue colors and a higher resistance to color loss compared to their nonacylated homologs. However, they remain vulnerable to a poorly understood combination of oxidative and hydrolytic reactions that strongly contribute to color loss and limits their industrial applications. In this work, the thermal degradation of isolated red cabbage anthocyanins (0, 1 or 2 acyl groups) at pH 7 was investigated by UPLC-DAD-MS (low- and high-resolution). Non-oxidative alterations, including deacylation and intramolecular acyl transfer, were observed and found very dependent on the number and position of the acyl group(s) as well as on the presence of iron ions. At intermediate and advanced thermal degradation, several oxidative mechanisms were evidenced that lead to protocatechuic acid, phloroglucinaldehyde 2-O-glucoside, acylglycosides and derivatives of 2,4,6-trihydroxyphenylacetic acid and 3,5,7-trihydroxycoumarin. Based on the product distribution observed and on the impact of added Fe²⁺ ions and H₂O₂, possible degradation mechanisms are discussed. They likely start with a one- or two-electron transfer from the anionic base (a major colored form in neutral solution) to O₂. The hydrogen peroxide produced could then further react as an electrophile with the anionic base and/or the hemiketal (major colorless hydrated form). This contribution to understanding the degradation mechanisms of anthocyanins around neutrality can open up new stabilization strategies to extend the range of their food applications to neutral media.

Full text of DOI:10.1016/j.dyepig.2020.108326

Dyes and Pigments 178 (2020) 108326
Contents lists available at ScienceDirect
Dyes and Pigments
journal homepage: http://www.elsevier.com/locate/dyepig
The fate of acylated anthocyanins in mildly heated neutral solution
,
,*
Julie-Anne Fengera **, Rebecca J. Robbinsb, Thomas M. Collins1, Olivier Danglesa
a Avignon University, INRAE, UMR408, 84000, Avignon, France
b Mars Wrigley, 1132 W Blackhawk Street, Chicago, IL, 60642, USA
A B S T R A C T
In neutral solution, anthocyanins acylated by hydroxycinnamic acids typically exhibit attractive blue colors and a higher resistance to color loss compared to their
nonacylated homologs. However, they remain vulnerable to a poorly understood combination of oxidative and hydrolytic reactions that strongly contribute to color
loss and limits their industrial applications. In this work, the thermal degradation of isolated red cabbage anthocyanins (0, 1 or 2 acyl groups) at pH 7 was
investigated by UPLC-DAD-MS (low- and high-resolution). Non-oxidative alterations, including deacylation and intramolecular acyl transfer, were observed and
found very dependent on the number and position of the acyl group(s) as well as on the presence of iron ions. At intermediate and advanced thermal degradation,
several oxidative mechanisms were evidenced that lead to protocatechuic acid, phloroglucinaldehyde 2-O-glucoside, acylglycosides and derivatives of 2,4,6-trihy-
droxyphenylacetic acid and 3,5,7-trihydroxycoumarin. Based on the product distribution observed and on the impact of added Fe2þ ions and H2O2, possible
degradation mechanisms are discussed. They likely start with a one- or two-electron transfer from the anionic base (a major colored form in neutral solution) to O2.
The hydrogen peroxide produced could then further react as an electrophile with the anionic base and/or the hemiketal (major colorless hydrated form).
This contribution to understanding the degradation mechanisms of anthocyanins around neutrality can open up new stabilization strategies to extend the range of
their food applications to neutral media.
1. Introduction
interactions with the anthocyanidin chromophore (intramolecular
copigmentation þ self-association), thereby protecting it against water
addition [1]. By contrast, the rate of irreversible degradation at pH 7
barely depends on the acylation pattern in the case of red cabbage an-
thocyanins [6]. Indeed, diacylation results in a higher proportion of the
colored anionic base at pH 7, which is probably a much better electron
donor (thus more vulnerable to autoxidation) than the neutral colorless
forms more readily accumulated from the weakly acylated pigments.
The mechanisms of irreversible degradation have been mainly
investigated at acidic pH [8–13] and only a few studies were also carried
out at neutral pH in the presence of radical initiators, H2O2 or ascorbate
(a H2O2 generator by autoxidation) [14,15]. In the pH range 2–4, an-
thocyanins were reported to be degraded into B-ring þ C2 (e.g., proto-
catechuic acid) and A-ring þ C4 (e.g., phloroglucinaldehyde) fragments,
C3 being probably eliminated as CO2. The colorless species, in particular
the chalcones, were proposed to be intermediates in the irreversible
degradation of anthocyanins at acidic pH [8,13]. At pH 7–8, the elec-
trophilic flavylium ion (a diacid with pKa ~4 and 7) is in trace amounts
so that water addition is very slow [16]. Anthocyanins are thus mostly a
mixture of neutral/anionic bases, with low concentrations of hemiketal
and neutral/anionic cis- and trans-chalcones (pKa ~8) gradually
Anthocyanins are plant pigments expressing a wide array of red to
blue colors depending on pH and the presence of other species suscep-
tible to interact with the anthocyanidin chromophore, such as phenolic
compounds (copigments) and metal ions [1]. In particular, anthocyanins
and their complexes can express attractive blue colors around neutral pH
[2]. This is the case of 3-O-sophorosyl-5-O-glucosylcyanidin and peo-
nidin derivatives commonly found in purple vegetables (e.g., red cab-
bage) [2–4]. Most importantly, purple vegetables are typically rich in
anthocyanins acylated by p-hydroxycinnamic acids (HCA), which in-
crease color stability and participate in color diversification. However,
at neutral and mildly alkaline pHs, color loss remains relatively fast,
which is a serious hurdle to industrial development.
Color loss in anthocyanin solutions at pH > 2 is due to the reversible
water addition to the flavylium ion (with concomitant accumulation of a
colorless hemiketal and pale yellow cis- and trans-chalcones) and to a
combination of irreversible routes (hydrolytic and oxidative pathways)
typically resulting in cleavage of the C-ring [5]. The color stability of
diacylated anthocyanins is much higher than for non- or mono-acylated
anthocyanins [2,6,7]. Indeed, the HCA residues can develop π-stacking
* Corresponding author.
** Corresponding author.
E-mail addresses: julie-anne.fenger@univ-avignon.fr (J.-A. Fenger), olivier.dangles@univ-avignon.fr (O. Dangles).
1
Retired.
https://doi.org/10.1016/j.dyepig.2020.108326
Received 30 January 2020; Received in revised form 22 February 2020; Accepted 29 February 2020
Available online 10 March 2020
0143-7208/© 2020 Elsevier Ltd. All rights reserved.

J.-A. Fenger et al.
Dyes and Pigments 178 (2020) 108326
appearing [17].
over 8 h and at 24 and 72 h. The total concentration in residual pigment
was quantified with a UV–Vis spectrophotometer (Agilent 8453)
immediately after cooling and acidification to pH 1.0–1.5 (fast conver-
sion of the residual colored forms, hemiketal and cis-chalcone into the
flavylium ion) and 6–50 h later (additional conversion of the trans--
chalcone) (Fig. 2-SI).
At pH 7, more stable blue colors can be obtained with cyanidin de-
rivatives in the presence of metal ions such as Fe2þ, Fe3þ and Al3þ [18].
Upon metal binding, the anthocyanidin chromophore adopts a p-qui-
nonemethide structure, which unlike the flavylium ion does not undergo
water addition with concomitant color loss. In spite of this color stabi-
lization, irreversible degradation occurs over a prolonged storage.
Addition of Fe2þ was reported to be actually protective with diacylated
anthocyanins, which strongly bind iron (a possible consequence of the
2.2. Product identification and quantification
strong
π-stacking interactions), but unexpectedly deleterious for the
The acidified and stabilized samples were analyzed with an Acquity
UPLC (Waters Corporation, Milford, USA) equipped with a binary sol-
non- and monoacylated homologs, suggesting that iron leakage from the
corresponding less stable complexes results in a prooxidant effect [6].
Unraveling the degradation mechanisms of anthocyanins around
neutrality can open up specific stabilization strategies for anthocyanins
used as blue colors. Therefore, this study aims at identifying the main
products of irreversible degradation at pH 7. A set of 3 anthocyanins
from red cabbage (from non-to diacylated) was thermally degraded at
neutral pH, 50 �C. The role of O2, added Fe2þ and hydrogen peroxide in
these mechanisms was investigated. Based on the product distribution
observed and on the impact of added Fe2þ ions and H2O2, possible
degradation mechanisms are discussed.
vent delivery manager and a diode array detector (DAD). Samples (5 μL)
were injected onto an Acquity UPLC BEH C18 reversed phase column
(50 � 2.1 mm, 1.7
μ
m) set at 30 �C. Phase A (1% HCO2H in H2O) and B
(1% HCO2H in MeCN) were eluted at 0.4 mL/min. Unless otherwise
specified, all chromatograms are presented at 280 nm after 24 h at pH 7,
50 �C. Only peaks with S/N > 8 on the 280 nm chromatogram were
considered for identification. For P1 and P4, gradient 1 (%B: 0 min: 2%,
12 min: 24%, 14–15 min: 80%, 16–18 min: 2%) was used. For PA,
gradient 2 (%B: 0–2.5 min: 2%, 7 min: 8%, 10 min: 24%, 13 min: 80%,
14–16 min: 2%) enabled a better separation of the more polar degra-
dation products.
2. Materials & methods
The UPLC system was coupled with a ESI-Q-trap HCT Ultra (Bruker
Daltonics, Bremen, Germany) in ultrascan mode. The capillary voltage
was À 1.8 kV (positive mode) or 2.2 kV (negative mode) with a 80–1500
m/z scanning interval at a speed of 26 � 103 m/z sÀ 1. Desolvation was
conducted with N2 at 365 �C, 40 psi, 540 L hÀ 1. Cone voltage was 40 V,
and the fragmentation amplitude was 1.2 V.
2.1. Materials
Red cabbage anthocyanins were isolated from red cabbage by prepa-
ratory LC according to already published procedures [19]. The pigments
investigated in this work encompass a nonacylated anthocyanin (PA) and
its homologs with a p-coumaroyl (pC) residue (P1) and an additional
sinapoyl (Sp) residue (P4). PA: cyanidin-3-O-[Glc-2-O-Glc]-5-O-Glc, P1:
cyanidin-3-O-[(6-O-pC)-Glc-2-O-Glc]-5-O-Glc, and P4: cyanidin-3-O-
[(6-O-pC)-Glc-2-O-(2-O-Sp)-Glc]-5-O-Glc (Scheme 1). Stock solutions (5
mM)ofpigmentwerepreparedinaqueous0.01MHCl(metal-tracegrade).
Aqueous H2O2, FeSO4, 7H2O, NaH2PO4, 2H2O, Na2HPO4, 7H2O and the
following standards for LC quantification, cyanin (cyanidin-3,5-O-diglu-
coside), p-coumaric acid and sinapic acid, were all obtained from
Sigma-Aldrich (St Louis, MO, USA). HPLC-MS grade water was used in all
experiments.
For confirmation of raw formulae, three samples of P1 (t ¼ 0, 24 h,
24 h in the presence of Fe2þ) were also analyzed on a Waters Acquity
UPLC system coupled with a Waters Synapt G2-Si High Resolution Mass
Spectrometer (HRMS) equipped with an ESI source (Waters Co.). The
source and desolvation temperatures were set at 120 �C and 500 �C,
respectively. Desolvation was also conducted with N2 at 500 �C (40 psi)
at 800 L hÀ 1. The capillary and cone voltages were set at 0.8 kV and 5 V,
respectively. The scan range was m/z 50–1500 with a spectrum acqui-
sition every 0.2 s and a resolution of 4 � 104. Mass scale was corrected
during acquisition using leucine enkephalin (Sigma-Aldrich). Data were
acquired using the MassLynx™ (V4.2) software in continuum mode. The
m/z accuracy (Δ, in ppm) of the parent ions was calculated as the
relative difference to their expected monoisotopic ion.
For each pigment, the fraction of colored and colorless species at
equilibrium at pH 7 (Scheme 1-SI, Fig. 1-SI) was calculated using the
global acidity constant of the flavylium ion (formation of the neutral
base þ colorless forms) and the two stepwise acidity constants
(sequential formation of the neutral and anionic bases) [2].
Anthocyanin were quantified as cyanin equivalent with a correction
factor accounting for the differences in molar absorption coefficient at
λmax(Vis) between the pigments [19]. The monoacylsophorose com-
pounds and coumarin derivatives were quantified in HCA equivalent
and phloroglucinaldehyde (PGA) derivatives in PGA equivalent.
The thermal degradation of anthocyanins was performed at 50 �C in
a thermostated water bath protected from light, as already described
[6]. The initial anthocyanin concentration in the 0.01 M phosphate
buffer was 5 � 10À 5 M. Aliquots were taken up at regular time intervals
3. Results
At pH 7 and 50 �C, the color stability of diacylated anthocyanin P4
and its iron complex is much higher than that of its non- and mono-
acylated counterparts PA and P1 [6]. However, the global color loss is
the result of a reversible component featuring water addition to the
flavylium ion (and subsequent isomerization steps) and of an irrevers-
ible component of true oxidative degradation. The latter component can
be appreciated by reversing the former through reacidification to pH
1.0–1.5, so as to convert all colored and colorless forms into the flavy-
lium ion. Hence, the time dependence of the residual flavylium per-
centage solely reflects the rate of oxidative degradation occurring in
neutral solution (Fig. 2-SI). After 24 h, the percentage of residual fla-
vylium ion lies in the range 40–60% and is mostly independent of the
acylation pattern. Product identification was then carried out (Fig. 1,
Table 1).
Three types of products were distinguished: a) colorless species
resulting from reversible water addition (trans-chalcones), b) new pig-
ments, i.e. isomers and deacylation products (Fig. 2), c) products of
Scheme 1. Structures of the red cabbage anthocyanins studied in this work.
2

J.-A. Fenger et al.
Dyes and Pigments 178 (2020) 108326
Fig. 1. Chromatograms of the solutions of PA (gradient 2), P1 and P4 (gradient 1) after 24 h at pH 7, 50 �C (detection at 280 nm). P5 (cyanidin-3-O-[(6-O-Fl)-Glc-2-
O-(2-O-Sp)-Glc]-5-O-Glc) is a contaminant of the P4 sample. Its chalcone (Ct5) and its hydrolysis product P2 (cyanidin-3-O-[(6-O-pC)-Glc-2-O-Glc]-5-O-Glc) are
also detected.
oxidative degradation. In addition, the composition of P1 and P4 solu-
tions after degradation over 24 h was determined under different con-
heating at pH 7 were shifted by À 1 nm and À 6 nm respectively, while
the λmax of PA remained unchanged. The decrease in λmax is ascribed to
deacylation, at a rate corresponding to the fraction of deacylation
products (PA, PB, P1). P1 and PB are respectively formed upon loss of
the Sp and pC residues (Fig. 2). When P1 and PB are heated separately
under the same conditions, 37% PA is formed from PB after 24 h, vs. only
14% from P1. This suggests that the Sp residue (at C2-OH of Glc-2) is
more prone to hydrolysis than the pC residue (at C6-OH of Glc-1). In-
vestigations with sucrose acylated by fatty acids (pH 7–10) also
concluded that esters of primary alcohols are more resistant to saponi-
fication than esters of secondary alcohols [21]. Anthocyanin deacylation
was observed previously in red cabbage extracts. Over storage, a
decrease in the diacylated anthocyanins was compensated by an in-
crease in the non- and monoacylated ones [22]. The kinetic monitoring
shows that PB is formed from P4 over the first 4 h, and that isomers of
both P4 and PB are formed later (Fig. 2).
ditions: addition of Fe2þ
, addition of hydrogen peroxide, inert
atmosphere. Finally, degradation routes are proposed and discussed.
3.1. Hydration and the reversible accumulation of the trans-chalcone
The fractions of colored and colorless species at the hydration
equilibrium (Scheme 1-SI, Fig. 1-SI) are strongly dependent on the
acylation pattern. From the global (hydration included) and specific
(sequential proton transfers) acidity constants of the flavylium ion at 25
�C [2] it is estimated that PA is almost colorless at pH 7 (99% hemiketal
B þ chalcones). By contrast, the fraction of colored forms is higher for P1
(ca. 15%) and P4 (ca. 80% colored forms, of which 55% anionic base).
In heated samples (50 �C), the residual trans-chalcone (Ct) can be
detected by UPLC-DAD-MS when the analyses are performed rapidly
after acidification. Indeed, its conversion into the flavylium ion is
strongly retarded by the slow cis-trans isomerization. The PA trans-
Upon heating at pH 7 and 8, a major P4 isomer accumulates (Table 2,
Fig. 3-SI). It displays a lower λmax (À 4 nm) than P4. Two other isomers
are also detected, both remaining very minor. By contrast, P1 isomeri-
zation is marginal (<2%). The hypothesis of cis-trans isomerization of
the HCA residues can be ruled out, first because the samples were heated
in the dark, and second, because it would also have occurred with P1.
Hence, the P4 isomers are believed to form upon migration (trans-
esterification) of a HCA residue within the same Glc. As acyl migration
within P1 is negligible, it can be assumed that the labile acyl residue of
P4 is the sinapoyl residue at C20-OH. Similar phenomena were reported
for aliphatic esters of sucrose in alkaline aqueous solution [21] with a
clear trend of acyl groups to shift from secondary to primary positions.
Overall, our data demonstrate that the sinapoyl residue of P4 is more
sensitive to both hydrolysis and trans-esterification than the p-cou-
maroyl residue.
chalcone (CtA) was detected with λ
¼ 330 nm (Fig. 3-SI), close to the
max
malvidin-3,5-diGlc Ct (λmax ¼ 335 nm) [20]. After 1 h, the fraction of Ct
reaches ca. 29% for PA (Fig. 2-SI). By comparison, at pH 6, the Ct
fraction accumulated from the triacylated heavenly blue anthocyanin,
which has the same glycosidation pattern as the red cabbage pigments, is
32% [17].
3.2. Deacylation and intramolecular acyl transfer
The acyl groups undergo hydrolysis and intramolecular migration
(trans-esterification) at pH 7 and 8. After 24 h at pH 7, the total yield of
these anthocyanin derivatives amounts to 7% for P1 and 28% for P4
(Table 2). The hydrolysis of the protecting acyl moieties must reduce the
color stability [1].
The λmax values of P1 and P4 in the visible range after a 24 h period of
3

J.-A. Fenger et al.
Dyes and Pigments 178 (2020) 108326
Table 1
Selection of ions detected in the solutions of PA, P1 and P4 after 24 h at pH 7, 50 �C. Products numbered according to elution order with the following convention:
product p potentially present in all 3 samples, product p’ potentially present in P1 and P4 samples, product p” specifically present in P4 samples (see Fig. 1).
#
Rt
Proposal
λmax
(nm)
m/z (À )
Ion type (À )
MS2 (À )
MS3 (À )
(min)
PA: Cya-3-[O-Glc-2-O-Glc]-5-O-Glc (gradient 2)
1
0.67
C3-Glc
275
345
[M-H]-
345 -> 165 (-Glc-H2O); 139 (-Glc-CO2); 183
(-Glc)
–
2–3
1.07
Unid.
291 325
643
1253
109
715
[MþCl]- [2M-
H]-
1253 -> 1055; 1235; 893; 643 -> 527; 617
1055 -> 893; 783; 551 527 -> 445; 783; 365
4
5
1.60
3.25
C2
258
325
[M-H-CO2]-
[MþCl]-
–
–
C1-3-Soph -5-
Glc
715 -> 517 (-Cl-Glc); 679 (-Cl)
517 -> 247; 191; 337
6
3.65
3.95
4.55
4.55
C4-Glc
Unid.
289
293
275
278
315
361
529
597
[M-H]-
–
[M-H]-
[M-H]-
315 -> 153 (-Glc)
153 -> 125 (-CO)
7
–
–
–
8
C6-Soph
C7-Soph
529 -> 409, 205 (-Soph)
597 -> 272 (-Soph-H); 417 (-Glc-H2O); 297;
555 (-CH2CO)
9a
272 -> 231; 258; 175
10
9b
4.95
5.78
Unid.
270 430
278
757
597
–
757 -> 551 (-Glc-CO2); 595 (-Glc); 713
(-CO2)
551 -> 371 (-Glc-H2O); 227; 281
272 -> 231; 258; 175
C7-Soph
[M-H]-
597 -> 272 (-Soph-H); 417 (-Glc-H2O); 297;
555 (-CH2CO)
CtA
PA
5.15
PA Ct
PA
330
510
789
[M-H]-
789 -> 627; 517; 285
627 -> 285; 517; 241
5.8–6.3
807 825
[MþCl-H]- Id.þ
H2O
807 -> 771; 609 825 -> 789; 627
771 -> 609; 285; 447 789 -> 627; 517; 285;
447
11
12
7.20
7.88
C3-Glc -Soph-C2
C3-Glc-C2
269
ND
805
481
[M-H]-
805 -> 463 (-Soph-H2O) 651 (-C2-H2O); 327
(-C2-Soph-H2O)
463 -> 327 (-C2); 299 (-pC-H2O); 165 (-C2-
Glc)
[M-H]-
481 -> 345 (-C2); 327 (-C2-H2O); 463
(-H2O)
345 -> 165 (-Glc-H2O); 139 (-Glc-CO2); 183
(-Glc); 327 (-H2O)
13
14
8.47
8.50
Cya-5-Glc
C5-Glc
ND
447
463
[M-2H]-
[M-H]-
447 -> 285 (-Glc)
285 -> 257
328
463 -> 419; 257; 445
419 -> 213
P1: Cya-3-O-[(6-O-pC)-Glc-2-O-Glc]-5-O-Glc (gradient 1)
10a
10b
10c
10d
20
2.53
3.44
3.92
4.55
5.21
5.80
pC-Soph
pC-Soph
pC-Soph
pC-Soph
C1-3-Soph
Unid.
310
310
310
310
310
ND
487
487
487
487
553
498
[M-H]-
[M-H]-
[M-H]-
[M-H]-
[MþCl]-
–
487 -> 469 (-H2O)
469 -> 205; 307 (-Glc); 265
487 -> 469 (-H2O)
469 -> 205; 307 (-Glc); 265
487 -> 469 (-H2O)
469 -> 205; 307 (-Glc); 265
487 -> 469 (-H2O)
469 -> 205; 307 (-Glc); 265
553 -> 499 (-Cl-H-H2O); 517 (-Cl-H)
498 -> 301 (Cya þ O); 336; 463 (Cya-Glc þ
O)
517 -> 499 (-H2O); 235; 295; 179
301 -> 165 (-C4 or C2); 257 (-CO2); 137
30
Ct1
40
5.97
7.45
P1 Ct
310
317
935 971
825
[M-H]- [MþCl]-
935 -> 773; 755; 663; 447; 285; 971 -> 935
825 -> 663 (-Glc)
663 -> 517; 247; 935 -> 773; 755; 663; 285
663 -> 517 (-pC); 247; 191
C1-3-(pC)Soph-
5-Glc
[M-H]-
P1
50
7.81
8.27
P1
524
308
917
951
[M-2H]-
[M-H]-
917 -> 755 (-Glc)
755 -> 609 (-pC); 339; 284; 309
C3-(pC)Soph-
Glc-C2
951 -> 623 (-C3-Glc); 463 (-pC-Soph-H2O)
463 -> 327 (-C2); 301 (-Glc); 165 (-C2-Glc)
60a
70
8.40
8.83
C7-(pC)Soph
C6-(pC)Soph
318
ND
743
675
[M-H]-
[M-H]-
743 -> 659; 597 (-pC); 272 (-pC-Soph-H)
675 -> 529 (-pC); 409; 205 (-pC-Soph)
659 -> 479 (-Glc-H2O); 335
529 -> 511 (-H2O); 409; 349 (-Glc-H2O); 205
(-Soph)
60b
9.30
C7-(pC)Soph
318
743
[M-H]-
743 -> 659; 597 (-pC); 272 (-pC-Soph-H)
659 -> 479 (-Glc-H2O); 335
P4: Cya-3-O-[(6-O-pC)-Glc-2-O-(2-O-Sp)-Glc]-5-O-Glc (gradient 1)
PB
1’’
P2
4.46
4.76
5.76
5.88
8.12
PB
530
309
530
530
530
1013
119
[MþCl-H]-
[M-H-CO2]-
[MþCl-H]-
[MþCl-H]-
[MþCl-H]- [M-
2H]-
1013 -> 977 (-Cl); 815 (-Glc)
–
983 -> 947 (-Cl); 785 (-Glc)
1013 -> 977 (-Cl); 815 (-Cl-Glc)
1159 -> 1123 (-Cl) 1123 -> 961 (-Glc)
977 -> 609 (-Glc)
pC acid
P2 (cont.)
PB isomer
P4 isomer
–
983
947 -> 785 (-Glc)
PBi
P4i
1013
1159
1123
1141
1123
1189
1067
977 -> 609 (-Sp-Glc); 339
961 -> 755 (-Sp); 737 (-Sp-H2O); 285 (Cya)
Ct4
P4
8.34
8.71
8.89
9.19
P4 Ct
320
534
534
ND
[M-H]-
1141 -> 977 (-pC-H2O); 869
1123 -> 961 (-Glc)
977 -> 853; 935 (-Sp); 469; 285 (Cya)
961 -> 755 (-Sp); 737 (-Sp-H2O); 285 (Cya)
1153 -> 991 (-Glc); 785 (-Glc-Sp); 947 (-Sp)
1031 -> 869 (-Glc); 663 (-Glc-Sp); 825 (-Sp);
517 (-Sp-pC-Glc)
P4
[M-2H]-
P5
P5 (cont.)
[MþCl-H]-
[MþCl]-
1189 -> 1153 (-Cl)
2’’
C1-Glc-(Sp,pC)
Soph
1067 -> 1031 (-Cl)
Ct4i
3’’a
3’’b
3’’c
9.31
10.0
10.6
11.1
P4 Ct isomer
C7-(Sp,pC) Soph
C7-(Sp,pC) Soph
C7-(Sp,pC) Soph
320
ND
ND
ND
1141
949
949
949
[M-H]-
[M-H]-
[M-H]-
[M-H]-
1141 -> 977 (-pC-H2O); 869; 855
949 -> 865; 931; 782
869 -> 663 (-Sp); 715; 645 (-Sp-H2O); 503
931 -> 725; 515; 359
949 -> 865; 931; 782
865 -> 711; 847; 539
949 -> 865; 931; 782
931 -> 545; 739; 311
3.3. The oxidative products & degradation routes
3.3.1. Confirmed structures
Compound 4 is detected in PA, P1 and P4 solutions. Its characteris-
tics are identical to those of a commercial standard of protocatechuic
acid (noted C2). Besides, C2 formation upon degradation of cyanidin
derivatives was reported several times [8,11]. Compound 4 is therefore
confidently identified as protocatechuic acid.
The products of irreversible degradation of PA, P1 and P4 were
characterized by UPLC-MS-DAD (Table 1). Several groups of compounds
only differ by the presence of the acyl and/or glucose moieties, and
share common fragments in MS2 and MS3. In this case, a common core
was assumed and tentatively identified (Scheme 2).
Compound 2’’ is detected in P4 solution only. The commercial
standard of p-coumaric acid (pC) displays the same characteristics. In
HRMS, the [M-H]- ion at m/z 163.0401 is also detected (Δ ¼ 3.6 ppm).
4

J.-A. Fenger et al.
Dyes and Pigments 178 (2020) 108326
Fig. 2. A: Kinetic monitoring of deacylation and intramolecular acyl transfer for diacylated pigment P4 (pH 7, 50 �C, insert for easier visual appreciation). The
difference between P4 consumption and accumulation of the neoformed pigments is ascribed to oxidative degradation. B: Kinetic scheme and hypothetical intra-
molecular acyl transfer routes within the sophorose moiety (Glc-2).
Table 2
Quantification by UPLC-DAD of anthocyanin degradation and of the new pigments formed in PA, P1 and P4 solutions after thermal treatment (24 h, pH 7, 50 �C) in the
absence or presence of Fe2þ. Concentrations in
μ
M of cyanin equivalent (corrected for differences in molar absorption coefficient between pigments).
Pigment
PA
Pigment þ Fe2þ (0.6 equiv.)
P1
P4
PA
P1
P4
t ¼ 0
t ¼ 24 h
t ¼ 0
t ¼ 24 h
t ¼ 0
t ¼ 24 h
t ¼ 0
51.4
t ¼ 24 h
t ¼ 0
2.4
t ¼ 24 h
t ¼ 0
t ¼ 24 h
PA
51.5
12.7
2.5
3.0
1.5
0.7
PB
0.8
1.4
1.4
3.6
2.3
3.8
0.5
1.1
P1
54.9
25.4
0.5
56.5
2.8
P1 isomer
P4 isomer
P4
0.4
9.1
8.1
1.3
0.6
51.7
53.4
13.5
Total anthocyanins
51.5
12.7
57.4
4.6
28.9
54.2
22.2
51.4
1.5
58.9
4.3
3.4
1.2
60.8
15.7
Acyl loss (%)
5.2
0.6
4.0
0.8
8.9
10.7
1.4
2.6
0.5
Acyl transfer (%)
16.8
Compound 2”, which is expected from the hydrolysis of P4 into PB, is
therefore confidently identified as p-coumaric acid.
solutions (Fig. 4-SI). They are thought to be derivatives of 3,5,7-trihy-
droxycoumarin (noted C1). The λ
of 5 (329 nm) is in agreement
with that of the 3,5-O-diglucoside mdaexrivative of C1 formed upon treat-
ment of malvidin 3,5-di-O-glucoside (malvin) by H2O2 in neutral solu-
tion [23]. As for 4′, its λmax (317 nm) is close to that of free p-coumaric
3.3.2. Probable structures
Compounds 5, 40 and 2” were respectively detected in PA, P1 and P4
5

J.-A. Fenger et al.
Dyes and Pigments 178 (2020) 108326
Scheme 2. Proposed core structures for the major compounds detected (see Table 1). C1 ¼ 3,5,7-trihydroxycoumarin, C2 ¼ protocatechuic acid, C3 ¼ 2,4,6-tri-
hydroxyphenylacetic acid, C4 ¼ phloroglucinaldehyde. Core structures C6 (MM ¼ 205) and C7 (MM ¼ 274) remain unidentified.
acid (310 nm) and in agreement with the reported 3-O-(p-coumaroyl)
are likely a mixture of regioisomers produced by migration of pC to a
glucoside of C1 (λ
¼ 315 nm) [23]. In HRMS, 4’ is detected as [M-H]-
neighboring OH group, each potentially present as a mixture of
anomers (Fig. 5-SI).
α and β
at m/z 825.2071,minaxagreement with the calculated monoisotopic value
(Δ ¼ 2.2 ppm). Among the fragments identified: the successive losses of
Glc to m/z 663.1552, and of p-coumaroylsophorose down to m/z
193.0127. The nonacylated coumarin (5) is also detected from P1 after
an extended thermal treatment of 96 h. Its raw formula is confirmed by
HRMS: m/z 715.1473 ([MþCl]-) and 679.1707 ([M-H]-, Δ ¼ 2.2 ppm).
The fragment at m/z 553 (from [MþCl]-) likely results from the loss of
Glc at C5-OH. By analogy, 2” must be the diacylated coumarin C1-3-(pC,
Sp)Soph-5-Glc. The following fragments substantiate this hypothesis:
869 ([M-H-Glc]-), 825 ([M-H-Sp]-), 663 ([M-H-Glc-Sp]-) and 517
([M-H-Glc-Sp-pC]-). Surprisingly, coumarin derivatives were not detec-
ted from anthocyanins that are not glycosylated at C5-OH [23]. In the
absence of added H2O2, these products remain in low amounts (<1% of
the initial pigment concentration, Table 1-SI).
As intramolecular acyl migration is negligible for P1, the pC moiety
appears labile in the cleavage products only. Thus, it seems that the
acyl–cyanidin
π-stacking interactions developed by P1 inhibit acyl
migration within the sophorose moiety. Only one diacylsophorose (m/z
693) in low concentration (<0.5 M) could be detected from P4, in
μ
agreement to the relatively high sensitivity of the sinapoyl residue to
hydrolysis. However, free sinapic acid remains undetected and must be
quickly consumed, while free p-coumaric acid is detected in P4 solution
(Table 1-SI). Consistently, under the same conditions, free sinapic acid
undergoes extensive oxidative dimerization after 24 h (22% residual
content, unpublished data) while p-coumaric acid is much more stable
(78% residual content).
From a 50
μ
M pigment solution, the pC-sophorose concentration
A series of compounds having a m/z of þ34 compared to the native
pigments were detected (Fig. 3). The compounds, noted 11, 12 and 5′,
display similar UV spectra and produce common fragments at m/z 463,
345, 327 and 301. Compound 11, detected from the 3 pigments, is
proposed to be C3(Glc,Soph)-C2, an analog of structures formed upon
reacting anthocyanins with H2O2 [15] or upon their azo-initiated
autoxidation [24]. Alternatively, a two-electron oxidized analog was
identified in the autoxidation of malvidin 3-O-glucoside in acidic solu-
tion [23,25]. From the [M-H]- ion of compound 12 (C3(Glc)-C2), the loss
of C2 and/or H2O followed by the loss of Glc and/or H2O or CO2, was
observed (Scheme 2-SI). The p-coumaroyl analog of compound 11 (5′,
m/z 951) yields fragments at m/z 623 (loss of C3 þ Glc) and 463 (loss of
pC þ Soph þ H2O). Second fragmentations of the latter ion give frag-
ments with m/z 327 (loss of C2), 301 (loss of Glc) and 165 (loss of C2 þ
Glc). In HRMS, 5’ is detected at m/z 951.2438 (C42H47O25, Δ ¼ 3.3
ppm). Finally, 12 (m/z 481) is identified as C3(Glc)-C2. Overall, these
compounds are proposed to be (acyl)glycosides of 2-(3,4-dihydrox-
y)-benzoyloxy-4,6-dihydroxyphenylacetic acid.
after a 72 h heating reached 11.2
μ
M and 6 M from P1 and P4,
μ
respectively. The concentration did not plateau, suggesting a high sta-
bility of these compounds. Overall, acylglycosides come up as useful
indicators of the oxidative degradation of acylated anthocyanins in
neutral solution.
Compound 6 is detected with the 3 pigments. Its main fragment (m/z
153) reflects the loss of glucose. The additional loss of 28 (m/z 125) is a
decarbonylation step expected for aldehydes. A probable structure for 6
is phloroglucinaldehyde 2-O-glucoside. The formula is in agreement
with the detected molecular ion at m/z 315.0714 (Δ ¼ 0.7 ppm) and the
aglycone at m/z 153.0203. Phloroglucinaldehyde and its glucoside were
frequently reported as anthocyanin degradation products involving the
A-ring [8,11,27]. Compound 6 accounts for ca. 10% of all products
present at 24 h in PA, P1 and P4 solutions (Table 1-SI).
3.3.3. Tentative structures
For the following compounds, no literature data is available. How-
ever, the structures proposed are compatible with at least two features
among raw formula from high resolution MS, MS2 fragments and
UV–visible spectrum.
Several isomers of compound 1’ (m/z 487) were detected from P1
(Fig. 5-SI). The two major ones 1′c and 1′d (m/z 487.1454 in HRMS,
corresponding to C21H27O13, Δ ¼ 0.5 ppm) were also detected from P4.
They mostly fragment by losing H2O. In MS3, the additional loss of Glc is
observed yielding a fragment ion at m/z 307. In ESI(þ), ions of m/z 511
and 527, respectively corresponding to the Naþ and Kþ adducts, were
detected. Compound 1′ is proposed to be pC-sophorose released by
hydrolysis of 1-O-acylglycosides formed during oxidative degradation
pathways. Four diacylsophoroses from purple sweet potato anthocya-
nins were reported and their structures confirmed by NMR [26]. No MS
fragmentation data have been reported yet. The pC-sophorose isomers
Compound 1 is detected with the 3 pigments (Fig. 6-SI). The [M-H]-
ion undergoes the loss of Glc but also the concomitant loss of Glc þ H2O,
with subsequent decarboxylation. This fragmentation pattern is close to
that of C3(Glc)-C2 (12, Scheme 2-SI). Based on these characteristics, 1 is
proposed to be 2-glucosyloxy-4,6-dihydroxyphenylacetic acid (C3Glc).
Compound 1 is probably produced by hydrolysis of the C3-C2 de-
rivatives identified above (11 and 12). HRMS confirmed the raw for-
mula proposed for 1: C14H18O10 (m/z 345.0810, Δ ¼ 3.4 ppm). This
group of products is mostly detected from the nonacylated anthocyanin
6

J.-A. Fenger et al.
Dyes and Pigments 178 (2020) 108326
Fig. 3. UV–visible, MS and MS2 spectra of a) C3(Glc)-C2 from PA (12, m/z 481); b) C3(Soph,Glc)-C2 from PA (11, m/z 805); c) C3(pCSoph,Glc)-C2 from P1 (50, m/
z 951).
(6.6% of the initial pigment concentration at 24 h, Table 1-SI).
reaction of cyanidin 3-O-glucoside with H2O2 in a water/ethanol
mixture [15]..
3.3.4. Other compounds
Compounds 8 and 7′ are sophorosides of the same unidentified
aglycone noted C6 (m/z 205.0131 for C10H5O5, Δ ¼ 2.9 ppm, Fig. 8-SI).
The chemical formulas of 8 (m/z 529.1190, Δ ¼ 0.7 ppm) and its p-
coumaroyl ester 7’ (m/z 675.1564, Δ ¼ 0.4 ppm) are respectively
Compound 14 is a minor product detected in PA solution. Its frag-
mentation pattern mainly consists in the loss of CO2 and water (Fig. 7-
SI). Compound 14 is identified to a glucoside of core structure C5
(Scheme 2). A closely related structure was reported previously in the
C22H26O15 and C31H32O17 .
7

J.-A. Fenger et al.
Dyes and Pigments 178 (2020) 108326
Similarly, compounds 9, 6′ and 3” are sophorosides of the same
structure noted C7 (m/z 272.0323, even value detected as a fragment
with both mass spectrometers, Fig. 9-SI), which has no equivalent in the
literature. The raw formula of C7 (C14H10O6) is compatible with the ions
detected for 6′ and 9 at the respective m/z values of 743.1816 and
597.1461 (Δ ¼ 1.0 and 0.9 ppm). C7 derivatives could be produced by a
multistep mechanism starting with the electrophilic addition of H2O2 to
the anionic base at position C3. Compounds 9, 6′ and 3” are all detected
as mixtures of 2 or 3 isomers. While acyl transfer can be proposed for 6’
and 3” to account for this observation, the isomerization of 9 remains
unexplained. Moreover, the absence of glucose at C5-OH, which is
normally not labile in neutral solution, is surprising. Hence, the struc-
ture proposed in Fig. 9-SI must be regarded as tentative.
4. Discussion
At neutral pH, anthocyanins are a mixture of neutral and anionic
bases slowly evolving into a mixture of hemiketal and chalcones. Upon
moderate heating in neutral solution, red cabbage anthocyanins evolve
by acyl hydrolysis and intramolecular transfer. The migration of the
sinapoyl group (at C2-OH of Glc-2) appears specific to an acyl residue
borne by a secondary C-atom. It is proposed to shift to the primary C-
atom (C6-OH, major isomer) through the 2 intermediate secondary C-
atoms (C3-OH and C4-OH, minor isomers). As most acylated anthocy-
anins display their acyl groups at primary C-atoms, this type of isom-
erization is generally not observed and constitutes a remarkable feature
of red cabbage anthocyanins. Interestingly, when these anthocyanins are
bound to iron, the sinapoyl residue loses its mobility. The well-known
propensity of HCA residues for developing π-stacking interactions with
3.4. Medium effects
the anthocyanidin nucleus [1,2] could be intensified within these
complexes, given the capacity of iron to coordinate up to 3 anthocyanin
ligands [28], thereby increasing the rigidity of the HCA residues and
inhibiting their migration.
The major products - other than anthocyanins - detected after 24 h in
PA, P1 and P4 are quantified in Table 1-SI. Besides the products of acyl
migration, protocatechuic acid (C2) and phloroglucinaldehyde-2-
glucoside (C4-Glc) come up as major products. The putative C6 and
C7 derivatives are also relatively abundant (ca. 10% of the initial
pigment concentration).
For red cabbage anthocyanins, the rate of anthocyanin consumption
(oxidative degradation) in neutral solution is not significantly different
for the di- and monoacylated pigments, and unexpectedly slightly faster
than for the nonacylated one [6]. This observation was interpreted by
assuming that PA is rapidly converted into the colorless forms (by
reversible water addition), which are much more resistant to autoxi-
dation than the electron-rich anionic base (far more abundant in solu-
tions of acylated anthocyanins) [6].
Fe2þ prevents the accumulation of the trans-chalcones through the
formation of metal complexes resistant to water addition. More sur-
prising is the almost total inhibition of P4 isomerization and deacyla-
tion. Higher concentrations of oxidation products, e.g. C7 derivatives 6’,
were detected in Fe2þ-supplemented P1 solutions (Fig. 10-SI) in agree-
ment with Fe2þ promoting P1 autoxidation [6]. This trend is not
observed with P4.
Upon degradation of an extract of purple sweet potato containing
acylated 3-O-sophorosyl-5-O-glucosylpeonidins (caffeoyl, feruloyl and
p-hydroxybenzoyl residues), the monoacylated anthocyanins appeared
more stable than the diacylated ones [29]. However, part of this
apparent stability could be due to the partial hydrolysis of diacylated
anthocyanins, thereby replenishing the pool of monoacylated
anthocyanins.
Addition of H2O2 (1 equiv.) to P1 solution leads to a much higher
concentration of pC-sophorose and coumarin derivatives and C3-C2
derivatives (Fig. 3). Addition of H2O2 in large excess (103 equiv.) in-
duces a fast consumption of the anthocyanin even in the absence of
thermal treatment. Under both conditions, pC-sophorose and C1 de-
rivatives are the major products. Moreover, a major, yet unidentified,
product (m/z 625 and 312, fragment at 183 corresponding to C3) is
specifically formed (Fig. 11-SI).
We recently showed that Fe2þ addition strongly slows down the rate
of color loss in P4 solution at pH 7, mostly because the p-quinoneme-
thide structure of P4 in the complex does not undergo water addition
[6]. This is consistent with Fe2þ addition inhibiting the formation of the
trans-chalcone. Besides its strong influence on the reversible color loss,
Fe2þ addition caused a modest slowing down of the early stage (up to 10
h at pH 7, 50 �C) of irreversible degradation for P4, while the opposite
holds for PA and P1 [6]. This difference was ascribed to the higher
stability of the iron – P4 (vs. iron – P1) complex due to enhanced
Under argon atmosphere (low O2 level), more residual pigment is
present after 24 h and the known oxidation products of P1 and PA are
very minor (Fig. 4).
Finally, in order to identify late degradation products of anthocya-
nins, the heating period was extended to 72 h. The chromatograms
(Fig. 12-SI) show the accumulation of protocatechuic acid from all three
pigments, and of the pC-sophorose isomers and coumarin p-coumar-
oylglycoside from P1 and P4 (as after addition of 1 equiv. H2O2).
Interestingly, none of the P4 degradation products bears the sinapoyl
residue (except traces of diacylsophorose). Again, the Sp residue is not
only more prone to intramolecular migration than the pC residue, but
also more labile or more reactive.
π
-stacking interactions, while leakage of iron from the iron – P1 complex
probably accelerates autoxidation. However, 24 h after iron addition, no
protection of P4 against irreversible degradation could be evidenced
(Table 2). On the other hand, the accumulation of oxidation products in
iron-supplemented solutions obviously remains more modest in P4 than
in P1 solution (Fig. 10-SI).
Our recent kinetic analysis suggests that the colored forms are pri-
marily involved in the oxidative degradation at pH 7 [6], which is
consistent with the anionic base being probably a much better electron
donor than the other (neutral) species. We thus assume that the first step
consists in an electron transfer from AÀ to O2 under the mediation of
transition metal traces, most probably Fe2þ. The aryloxyl radical thus
formed can evolve through 2 distinct pathways (Scheme 3):
a) A second electron transfer to form a highly electrophilic o-quinone
intermediate (pathway specific to B-rings having a 30,40-dihydroxy
substitution such as cyanidin derivatives) with concomitant gener-
ation of H2O2. Then, the o-quinone is expected to add a water
molecule, thereby leading to intermediate I1.
b) Addition of O2 with formation of a highly reactive peroxyl radical,
which will rapidly abstract a labile H-atom from a second anthocy-
anin molecule, thus yielding intermediate I2, a hydroperoxide.
Fig. 4. Distribution of degradation products from P1 after 24 h at pH 7, 50 �C.
A: P1. B-C: Impact of added H2O2 (B: 1 equiv., C: 103 equiv.), D: Impact of an
argon atmosphere.
8

J.-A. Fenger et al.
Dyes and Pigments 178 (2020) 108326
Scheme 3. Proposed mechanisms for the early stages of anthocyanin autoxidation in neutral solution.
Intermediates I1 and I2 may have different fates, some leading to
products identified in this work or in the literature. In particular, I1 can
add a second water molecule and form an intermediate already postu-
lated to result from the electrophilic attack of H2O2 to the hemiketal in
acidic solution [15]. From this intermediate, two end-products
(belonging to the C3-C2 and C5 groups) duly identified by NMR can
be produced (Scheme 4). Alternatively, elimination of the glycosyl
group at C3-OH is feasible. More generally, the conversion of the
glycosidic bond at C3-OH into an ester bond opens up a route for the
release of the glycosyl group in neutral solution through simple
hydrolysis.
flavylium ion (C2 position) or chalcone (Bayer-Villiger reaction). The
first route has been convincingly demonstrated in acidic solution from
labelling experiments (reaction with H128O2 or in H218O) [15]. It leads to
intermediate I1 (also produced by two-electron oxidation and subse-
quent water addition, Scheme 3) or its water adduct. The second route
has the additional advantage to rationalize the formation of the
coumarin derivatives. On the one hand, these products are detected at
pH 5–7 but not at pH < 3 [23], which is not consistent with a mechanism
involving the flavylium ion. On the other hand, addition of H2O2 indeed
promotes their formation, e.g. 4’ (Fig. 4). Overall, the second route re-
mains possible, although coumarins might be also produced through
autoxidation of the anionic base (Scheme 3-SI). However, complemen-
tary products derived from the B-ring (p-hydroquinones in the
Bayer-Villiger rearrangement, p-quinones in the autoxidation route)
were not detected.
Similar mechanisms can be written from I2 (Scheme 5). In this case,
C3-C2 compounds are also expected, although in a two-electron
oxidized version. Such a compound (two (Z,E) isomers) was indeed
fully identified by NMR in the autoxidation of malvidin 3-O-glucoside in
acidic solution [25]. With a malvidin derivative (no catechol ring), the
two-electron oxidation pathway is quenched and O2 addition is actually
the most likely fate for the aryloxyl radical. However, with the cyanidin
derivatives investigated in this work, only the reduced version was
evidenced, an indication that the two-electron oxidation pathway is
privileged (Schemes 3 & 4) and/or that H2O2 addition to the anthocy-
anins also occurs (see below). Alternatively, formation of a 1,2-dioxe-
tane ring (with concomitant re-aromatization of the A-ring) might
open up a route for the formation of C4 derivatives (Scheme 5). Phlor-
oglucinaldehyde and its glycosides are actually classical markers of
anthocyanin degradation [10,11]. They could be formed by other
routes, such as H2O2 addition to C3ox (free acid), followed by decar-
boxylation, or retro-aldol condensation from C-ring-opened in-
termediates (Scheme 4).
Finally, no direct participation of the HCA residues in the oxidative
degradation could be evidenced and analyses by UPLC-DAD-MS and by
capillary zone electrophoresis failed to detect anthocyanin dimers or
higher oligomers.
5. Conclusions
Under the conditions where anthocyanins express blue colors, i.e. pH
7 in the presence of metal ions or pH 8, they undergo oxidative and
hydrolytic pathways that alter the color and restrict their applications.
The irreversible degradation of acylated red cabbage anthocyanins at 50
�C leads to several groups of products, among which
phloroglucinaldehyde-2-glucoside, p-coumaroylsophorose (a mixture of
regioisomers) and derivatives of 2-(3,4-dihydroxy)benzoyloxy-4,6-
dihydroxyphenylacetic acid are the major ones. Overall, the acylglyco-
sides (p-coumaroylsophorose in this work) appear particularly stable
and thus constitute suitable markers of the irreversible degradation of
acylated anthocyanins.
Hydrogen peroxide produced in the autoxidation step probably
participates in the oxidative degradation (as suggested by the experi-
ments with added H2O2), either by electrophilic attack onto the anionic
base or hemiketal (C3 position), or by nucleophilic attack onto the
9

J.-A. Fenger et al.
Dyes and Pigments 178 (2020) 108326
Scheme 4. Proposed mechanisms for the fate of intermediate I1.
In addition, the diacylated red cabbage anthocyanins appear
remarkably prone to isomerization by intramolecular acyl transfer, a
phenomenon that is evidenced for the first time.
Overall, the irreversible degradation of anthocyanins in neutral so-
lution is probably kinetically controlled by an initial step of one- or two-
electron autoxidation of the anionic base. The major oxidation products
are thus proposed to derive either from the oxidized anionic base itself
or from an electrophilic attack of H2O2 (produced in the autoxidation
step) to the anionic base.
For the development of anthocyanin extracts as food colorants in
neutral media, the priority should be set at providing protection against
autoxidation, for instance by the formation of stable redox-inert metal
complexes or by adding suitable antioxidants.
Declaration of competing interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
the work reported in this paper.
CRediT authorship contribution statement
Julie-Anne Fenger: Conceptualization, Methodology, Formal anal-
ysis, Investigation, Writing - original draft, Visualization. Rebecca J.
Robbins: Resources, Funding acquisition. Thomas M. Collins: Re-
sources, Funding acquisition. Olivier Dangles: Conceptualization,
Methodology, Writing - review & editing, Supervision.
Acknowledgements
Scheme 5. Proposed mechanisms for the fate of intermediate I2.
The HRMS analyses were carried out on the ESI-Q-trap of the
10

J.-A. Fenger et al.
Dyes and Pigments 178 (2020) 108326
degradation products. Food Chem 2017;214:234–41. https://doi.org/10.1016/j.
foodchem.2016.07.071.
Metaboscope Platform funded by the European Fund for Regional
Development, the French Ministry of Research, Higher Education and
[12] Stebbins NB. Ascorbic acid-catalyzed degradation of cyanidin-3- O-β-glucoside:
proposed mechanism and identification of a novel hydroxylated product. J Berry
Res 2016;6:175–87. https://doi.org/10.3233/JBR-160132.
^
Innovation, the Region Provence-Alpes-Cote d’Azur, the Departmental
Council of Vaucluse and the Urban Community of Avignon. The authors
gratefully thank Christian Ginies for his support in the MS analyses, and
[13] Satake R, Yanase E. Mechanistic studies of hydrogen-peroxide-mediated
anthocyanin oxidation. Tetrahedron 2018;74:6187–91. https://doi.org/10.1016/j.
tet.2018.09.012.
�
Dr. Veronique Cheynier for helpful discussion.
[14] Pina F. Chemical applications of anthocyanins and related compounds. A source of
bioinspiration. J Agric Food Chem 2014;62:6885–97. https://doi.org/10.1021/
jf404869m.
Appendix A. Supplementary data
[15] Mendoza J, Basílio N, Pina F, Kondo T, Yoshida K. Rationalizing the color in
heavenly blue anthocyanin. A complete kinetic and thermodynamic study. J Phys
Chem B 2018;122:4982–92. https://doi.org/10.1021/acs.jpcb.8b01136.
[16] Buchweitz M, Brauch J, Carle R, Kammerer DR. Colour and stability assessment of
blue ferric anthocyanin chelates in liquid pectin-stabilised model systems. Food
Chem 2013;138:2026–35. https://doi.org/10.1016/j.foodchem.2012.10.090.
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.dyepig.2020.108326.
References
[17] Ahmadiani N, Robbins RJ, Collins TM, Giusti MM. Molar absorptivity (ε) and
spectral characteristics of cyanidin-based anthocyanins from red cabbage. Food
Chem 2016;197:900–6. https://doi.org/10.1016/j.foodchem.2015.11.032.
[18] Preston NW, Timberlake CF. Separation of anthocyanin by high-performance liquid
chromatography. J Chromatogr 1981:222–8.
[1] Trouillas P, Sancho-García JC, De Freitas V, Gierschner J, Otyepka M, Dangles O.
Stabilizing and modulating color by copigmentation: insights from theory and
experiment. Chem Rev 2016;116:4937–82. https://doi.org/10.1021/acs.
chemrev.5b00507.
�
[19] Thevenet S, Wernicke A, Belniak S, Descotes G, Bouchu A, Queneau Y.
[2] Dangles O, Fenger J-A. The chemical reactivity of anthocyanins and its
consequences in food science and nutrition. Molecules 2018;23:1970. https://doi.
org/10.3390/molecules23081970.
Esterification of unprotected sucrose with acid chlorides in aqueous medium:
kinetic reactivity versus acyl- or alkyloxycarbonyl-group migrations. Carbohydr
Res 1999;318:52–66. https://doi.org/10.1016/S0008-6215(99)00079-8.
[20] Wiczkowski W, Szawara-Nowak D, Topolska J. Changes in the content and
composition of anthocyanins in red cabbage and its antioxidant capacity during
fermentation, storage and stewing. Food Chem 2015;167:115–23. https://doi.org/
10.1016/j.foodchem.2014.06.087.
[3] Moloney M, Robbins RJ, Collins TM, Kondo T, Yoshida K, Dangles O. Red cabbage
anthocyanins: the influence of d-glucose acylation by hydroxycinnamic acids on
their structural transformations in acidic to mildly alkaline conditions and on the
resulting color. Dyes Pigments 2018;158:342–52. https://doi.org/10.1016/j.
dyepig.2018.05.057.
[21] Hrazdina G, Franzese AJ. Oxidation products of acylated anthocyanins under
acidic and neutral conditions. Phytochemistry 1974;13:231–4. https://doi.org/
10.1016/S0031-9422(00)91300-1.
[4] Fenger J-A, Moloney M, Robbins RJ, Collins TM, Dangles O. The influence of
acylation, metal binding and natural antioxidants on the thermal stability of red
cabbage anthocyanins in neutral solution. Food Funct 2019;10:6740–51. https://
doi.org/10.1039/C9FO01884K.
[22] Kamiya H, Yanase E, Nakatsuka S. Novel oxidation products of cyanidin 3-O-
glucoside with 2,20-azobis-(2,4-dimethyl)valeronitrile and evaluation of
anthocyanin content and its oxidation in black rice. Food Chem 2014;155:221–6.
https://doi.org/10.1016/j.foodchem.2014.01.077.
[5] Sigurdson GT, Robbins RJ, Collins TM, Giusti MM. Effects of hydroxycinnamic
acids on blue color expression of cyanidin derivatives and their metal chelates.
Food Chem 2017;234:131–8. https://doi.org/10.1016/j.foodchem.2017.04.127.
[6] Cabrita L, Petrov V, Pina F. On the thermal degradation of anthocyanidins:
cyanidin. RSC Adv 2014;4:18939–44. https://doi.org/10.1039/C3RA47809B.
[7] Es-Safi N-E, Meudec E, Bouchut C, Fulcrand H, Ducrot P-H, Herbette G, et al. New
compounds obtained by evolution and oxidation of malvidin 3- O -glucoside in
ethanolic medium. J Agric Food Chem 2008;56:4584–91. https://doi.org/
10.1021/jf8001872.
[23] Lopes P, Richard T, Saucier C, Teissedre P-L, Monti J-P, Glories Y. Anthocyanone A:
A quinone methide derivative resulting from malvidin 3-O-glucoside degradation.
J Agric Food Chem 2007;55:2698–704. https://doi.org/10.1021/jf062875o.
[24] Terahara N, Matsui T, Minoda K, Nasu K, Kikuchi R, Fukui K, et al. Functional new
acylated sophoroses and deglucosylated anthocyanins in a fermented red vinegar.
J Agric Food Chem 2009;57:8331–8. https://doi.org/10.1021/jf901809p.
[25] Piffaut B, Kader F, Girardin M, Metche M. Comparative degradation pathways of
malvidin 3,5-diglucoside after enzymatic and thermal treatments. Food Chem
1994;50:115–20. https://doi.org/10.1016/0308-8146(94)90106-6.
[8] Fleschhut J, Kratzer F, Rechkemmer G, Kulling SE. Stability and biotransformation
of various dietary anthocyanins in vitro. Eur J Nutr 2006;45:7–18. https://doi.org/
10.1007/s00394-005-0557-8.
�
�
[26] Estevez L, Sanchez-Lozano M, Mosquera RA. Complexation of common metal
cations by cyanins: binding affinity and molecular structure. Int J Quant Chem
2019;119:1–11. https://doi.org/10.1002/qua.25834.
[9] Sadilova E, Carle R, Stintzing FC. Thermal degradation of anthocyanins and its
impact on color and in vitro antioxidant capacity. Mol Nutr Food Res 2007;51:
1461–71. https://doi.org/10.1002/mnfr.200700179.
[27] Xu J, Su X, Lim S, Griffin J, Carey E, Katz B, et al. Characterisation and stability of
anthocyanins in purple-fleshed sweet potato P40. Food Chem 2015;186:90–6.
https://doi.org/10.1016/j.foodchem.2014.08.123.
[10] Seeram NP, Bourquin LD, Nair MG. Degradation products of cyanidin glycosides
from Tart Cherries and their bioactivities. J Agric Food Chem 2001;49:4924–9.
https://doi.org/10.1021/jf0107508.
[11] Sinela A, Rawat N, Mertz C, Achir N, Fulcrand H, Dornier M. Anthocyanins
degradation during storage of Hibiscus sabdariffa extract and evolution of its
11