The fate of flavanol-anthocyanin adducts in wines: Study of their putative reaction patterns in the presence of acetaldehyde

Article abstract of DOI:10.1016/j.foodchem.2010.01.060

Methylmethine bridged flavanol-anthocyanin dimeric adducts (F-A) were synthesised by hemisynthesis using catechin, acetaldehyde and F-A adducts as precursors in order to simulate putative reactions undergoing in red wine. The products obtained, catechin-m

Full text of DOI:10.1016/j.foodchem.2010.01.060

Food Chemistry 121 (2010) 1129–1138
Contents lists available at ScienceDirect
Food Chemistry
journal homepage: www.elsevier.com/locate/foodchem
The fate of flavanol–anthocyanin adducts in wines: Study of their putative
reaction patterns in the presence of acetaldehyde
*
Frederico Nave, Natércia Teixeira, Nuno Mateus, Victor de Freitas
Centro de Investigação em Química, Departamento de Química, Faculdade de Ciências, Universidade do Porto, Rua do Campo Alegre 687, 4169-007 Porto, Portugal
a r t i c l e i n f o
a b s t r a c t
Article history:
Methylmethine bridged flavanol–anthocyanin dimeric adducts (F–A) were synthesised by hemisynthesis
using catechin, acetaldehyde and F–A adducts as precursors in order to simulate putative reactions
undergoing in red wine. The products obtained, catechin–methylmethine–catechin–malvidin-3-O-glu-
cose and methylmethine-(catechin–malvidin-3-O-glucose)2, were analysed by HPLC–ESI-MS. The frag-
mentation patterns allowed the confirmation of the structures. These compounds were investigated
and found in wine samples (2-year-old table wine and 10-year-old Port wine), thus confirming their for-
mation in vino as a result of flavanol–anthocyanin adducts transformations.
Received 19 October 2009
Received in revised form 19 November 2009
Accepted 20 January 2010
Keywords:
Flavanol–anthocyanin adducts
Wine pigments
Hemisynthesis
Ó 2010 Elsevier Ltd. All rights reserved.
Acetaldehyde
HPLC–ESI-MS
1. Introduction
Skouroumounis, & Hayasaka, 2004); (d) formation of a pyranic ring
as a result of the reaction between anthocyanins and molecules
Anthocyanins are phenolic compounds belonging to the group
of flavonoids and are responsible for plant pigmentation. They
are often found in the epidermic tissue of flowers and fruits (Al-
bearing a vinyl moiety, such as acetaldehyde (B-type vitisins) (Al-
calde-Eon, Escribano-Bailón, Santos-Buelga, Rivas-Gonzalo,
&
2004; Drinkine et al., 2007a), pyruvic acid (A-type vitisins), aceto-
acetic acid (He, Santos-Buelga, Silva, Mateus, & de Freitas, 2006),
vinylphenols or vinylflavonols (Alcalde-Eon et al., 2004); (e) alde-
hyde mediated condensation between anthocyanin and flavan-3-
ols (Alcalde-Eon et al., 2004, 2006; Drinkine et al., 2007a; Gonz-
alez-Paramas et al., 2006); (f) direct condensation between antho-
cyanin and flavan-3-ols (Duenas, Fulcrand, & Cheynier, 2005;
Remy, Fulcrand, Labarbe, Cheynier, & Moutounet, 2000; Salas
et al., 2004a).
The direct condensation between flavan-3-ols and anthocya-
nins may originate both anthocyanin–flavanol (A–F) and flava-
nol–anthocyanin (F–A) adducts, depending on the reaction
mechanism. These compounds were initially postulated by Jurd
and Sommers in the late 1960s (Alcalde-Eon et al., 2004; Gonz-
calde-Eon, Escribano-Bailón, Santos-Buelga,
& Rivas-Gonzalo,
2006; Borkowski, Szymusiak, Gliszczynska-Swiglo, & Tyrakowska,
2005; Es-Safi et al., 2008; Vidal, Cartalade, Souquet, Fulcrand, &
Cheynier, 2002). They occur as water soluble glycosides or acylgly-
cosides, the latter being aliphatic, phenolic or hydroxycinnamic
acid residues esterified to the sugar moieties (Borkowski et al.,
2005). During the winemaking process, anthocyanins are extracted
from grape berries, thus becoming directly responsible for the red
wine colour (Alcalde-Eon et al., 2006; Borkowski et al., 2005).
Other anthocyanin features such as antioxidant capacity, radical
scavenging and metal chelation ability may also play an important
role in wine quality (Vidal et al., 2002).
Monomeric anthocyanins do not remain stable for a long time
in wine (Brouillard, Chassaing, & Fougerousse, 2003). In fact, it is
believed that they undergo several types of reactions: (a) oxidation
(Es-Safi et al., 2008); (b) aldehyde mediated condensation between
anthocyanins (Abe et al., 2008; Drinkine, Lopes, Kennedy, Teisse-
dre, & Saucier, 2007a; Gonzalez-Paramas et al., 2006); (c) direct
condensation between anthocyanins (Alcalde-Eon, Escribano-Bai-
lon, Santos-Buelga, & Rivas-Gonzalo, 2007; Lopes-da-Silva, Escrib-
alez-Manzano et al., 2008) but they remained
a theoretical
hypothesis until the late 1990s/early 2000s when their presence
in red wines and chemical structures were demonstrated by
HPLC–MS (Vivar-Quintana, Santos-Buelga, Francia-Aricha,
&
Rivas-Gonzalo, 1999), thiolysis (Remy et al., 2000), hemisynthesis
(Salas et al., 2004a), mass fragmentation patterns (Gonzalez-Para-
mas et al., 2006; Salas et al., 2004a) and NMR (Gonzalez-Manzano
et al., 2008; Salas, Le Guerneve, Fulcrand, Poncet-Legrand, &
Cheynier, 2004b). In the last decade, this type of compounds
was also detected and identified in plants such as strawberries
(Fossen, Rayyan, & Andersen, 2004), beans, grapes (Gonzalez-
ano-Bailon,
& Santos-Buelga, 2007; Vidal, Meudec, Cheynier,
* Corresponding author.
E-mail address: vfreitas@fc.up.pt (V. de Freitas).
0308-8146/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.foodchem.2010.01.060

1130
F. Nave et al. / Food Chemistry 121 (2010) 1129–1138
Paramas et al., 2006) and corn (Gonzalez-Manzano et al., 2008;
Gonzalez-Paramas et al., 2006).
McGraw (1983) and consisted in a mild acidic cleavage of procy-
anidin in the presence of Mv3glc. In short, in a 50 ml schlenk under
argon atmosphere, 60 mg of Mv3glc were placed in 15 ml of 10%
formic acid aqueous solution, followed by 4 mg of procyanidin B4
and 30 ml of ethyl ether. After sealing, the schlenk was heated
and stirred for 3 h at 85 °C. At the end, the organic phase was dis-
carded. The addition of procyanidin and ethyl ether was repeated
four times. The aqueous phase was then fractionated as described
in Section 2.3.
The previous works demonstrated that the F–A adducts possess,
when compared to their anthocyanic precursors, different colour
properties (batochromic shift of 10 nm) but similar chemical stabil-
ity towards hydration and sulphite bleaching (Salas et al., 2004b).
Even though the in vitro hemisynthesis of Cat–Mv3glc adducts
became possible in 2004, no studies regarding their stability/reac-
tivity in wine-like solutions were published since then.
The F–A adducts are composed by one flavan-3-ol moiety
(upper unit) linked to an anthocyanin moiety (lower unit) by a
B-type C4–C8 interflavanic bond. For that reason they should
present typical reactivity of both flavanols and anthocyanins.
So, they are expected to react, as nucleophiles at the C6 and C8
carbons of the flavanol upper unit and as electrophiles at the C2
and C4 carbons of the anthocyanin lower unit. As the latter kind
of reactions were already regarded (Salas et al., 2004), the present
work tried to focus on the possible reactions occurring in the
upper unit of the adduct. Since red wines tend to have a high
content in acetaldehyde either by yeast production or ethanol
oxidation, the most probable reactions would be the formation
of methylmethine bridged adducts (Drinkine, Lopes, Kennedy,
Teissedre, & Saucier, 2007b). These methylmethine bridged com-
pounds would be anthocyanin–methylmethine–adduct, flavan-3-
ol-methylmethine–adduct and adduct–methylmethine–adduct.
In order to studynew putative wine pigments such as these, a pos-
sible strategy is to chemically synthesize them starting from selected
organic precursors. Structure elucidation may be subsequently con-
ducted by mass and NMR spectroscopies (Brouillard et al., 2003).
Taking this strategy into account, both catechin–methylme-
thine–catechin–Mv3glc and catechin–Mv3glc–methylmethine–
catechin–Mv3glc were synthesised, characterised (UV/vis and
mass spectroscopy) and investigated in real wine samples. This
work is the first report of the UV–vis spectra and mass fragmenta-
tion patterns of the products of the reaction between flavan-3-ol-
anthocyanin dimeric adducts, flavan-3-ols and acetaldehyde.
2.3. Purification of the adduct Cat–Mv3glc
The fractionation and purification of the synthesised adduct
was performed in three different stages:
2.3.1. Stage 1 – Rough fractionation
The evaporated aqueous phase of the procedure described in
Section 2.2. was directly applied to 50 g of Toyopearl HW40(s)
gel in a 10 cm diameter medium porosity sintered glass funnel
with connected to standard vacuum filtration glassware and grad-
ually eluted with increasing percentage (v/v) of acidified methanol
(20%, 80% and 100%).
2.3.2. Stage 2 – Refined fractionation
The 80% (v/v) acidic methanolic solution was further fraction-
ated by column (i.d. 16 mm) chromatography. The stationary
phase was composed by 30 g of Toyopearl HW40(s) and the mobile
phase by a series of 40% (v/v), 75% (v/v) and 100% (v/v) of acidic
methanol. The flow rate of the peristaltic pump was set at
0.8 ml minÀ1. The collection of the fractions was carried out by
the visual detection of the coloured bands formed.
2.3.3. Stage 3 – Purification by preparative reversed phase HPLC
The specifications are described in Section 2.6.3. The eluted
peaks were collected, frieze dried and kept away from the light
at À18 °C until further use.
2.4. Hemisynthesis of Cat–Mv3glc derived compounds
2. Materials and methods
2.4.1. Hemisynthesis of catechin–methylmethine–catechin-(4,8)–
malvidin-3-O-glucoside
2.1. Reagents
Cat–Mv3glc (0.32
solution were transferred to a 750
l
mol) and 500
l
l of 20% aqueous ethanolic
Organic solvents, ethyl ether (pure) and ethyl acetate (pure) and
methanol (pure) were purchased from José M. Vaz Pereira (Lisboa,
Portugal) and acetonitrile (HPLC grade) and ethanol (Absolute)
were purchased from Carlo Erba (Val de Reuil, France).
(+)-Catechin hydrate (p. 98%) and (À)-epicatechin (p. 98%) were
purchased from Sigma–Aldrich (Madrid, Spain). (+)-Taxifolin
(p. 98%) was purchased from Extrasynthèse (Genay, France).
Malvidin-3-O-glucoside (Mv3glc) was extracted and purified
from a young red wine (c.v. Touriga Nacional) by semi-preparative
HPLC with C18 reversed phase column, as described elsewhere (Pis-
sarra, Mateus, Rivas-Gonzalo, Buelga, & de Freitas, 2003).
Sodium borohydride (p. 98%) was purchased from Sigma–Al-
drich (Madrid, Spain). Acetaldehyde (p. 99%) and formic acid MS
grade were obtained from Fluka (Buchs, Switzerland). Formic acid
(p. 98%) and hydrocloric acid (37% v/v) were purchased from Pan-
reac (Barcelona, Spain).
l
l microtube. The pH was ad-
justed to 3.0 with diluted hydroalcoholic HCl. It was then added
1.0 and 10.0 M equivalents of (+)-catechin and acetaldehyde,
respectively. The microtube was sealed and incubated at 35 °C
for 10 days. The reaction was followed both by HPLC–DAD and
HPLC–MS.
2.4.2. Hemisynthesis of methylmethine-(catechin-(4,8)–malvidin-3-
O-glucoside)2
The procedure used for this synthesis was the same as the
method described in the Section 2.4.1. without (+)-catechin.
2.5. Wine sample fractionation
Two different kinds of red wines were used to investigate the
presence of the methylmethine bridged Cat–Mv3glc derivatives
in wine: a 2-year-old red wine from the Douro Demarcated Region
(Portugal) and a 10-year-old Vintage Port wine.
Procyanidin B4 dimer was hemisynthesised according to the
procedure described by Pissarra et al. (2004).
Both wine samples were processed similarly. Briefly, 750 ml of
wine were dealcoholised in a rotary evaporator (T ꢀ 35 °C), applied
to a reversed phase C18 placed into a 10 cm diameter medium
porosity sintered glass funnel connected to standard vacuum filtra-
tion glassware, washed with water to remove the sugars, salts and
other organic acids and finally eluted with acidic methanol to re-
2.2. Hemisynthesis of the adduct catechin-(4,8)–malvidin-3-O-
glucoside (Cat–Mv3glc)
The hemisynthesis of the adducts Cat–Mv3glc was partially
based on the works of Haslam (1980) and Hemingway and

F. Nave et al. / Food Chemistry 121 (2010) 1129–1138
1131
move the phenolic compounds. The methanol was evaporated un-
der vacuum in the presence of water (250 ml) and the remaining
solution extracted with ethyl acetate (3 Â 250 ml). The organic
phases were discarded and the aqueous phases containing the
red wine pigments were fractionated with 50 g of Toyopearl
HW40(s) in a sintered glass funnel under vacuum and consecu-
tively eluted with 200 ml of acidified water and 20% (v/v), 40%
(v/v), 60% (v/v), 80% (v/v) and 100% (v/v) of aqueous acidic metha-
nol. The fractions were concentrated, in a rotary evaporator, up to
10 ml of water and analysed by HPLC–MS prior to freeze-drying.
and 30% (v/v) methanol. The rate flow used was 4 ml minÀ1. The
gradient method started with a linear gradient ranging from 20%
B to 66% of B in 50 min followed by another linear gradient from
66% to 100% of B in 10 min. At that point, an isocratic step was
maintained at 100% B for another 10 min proceeded by a re-equil-
ibration step of 15 min of 20% of B.
3. Results and discussion
3.1. Hemisynthesis of Cat–Mv3glc adduct
2.6. Chromatographic conditions
The adducts Cat–Mv3glc were synthesised through a nucleo-
philic addition of a hemiketalic malvidin-3-glucoside to a catechin
C4 carbocation. The carbocation was obtained by a mild acidic
cleavage of a dimeric procyanidin (B4). Salas et al. (2004a) have
previously proposed a very similar hemisynthetic approach to
synthesize this adduct (Salas, Fulcrand, Meudec, & Cheynier,
2003; Salas et al., 2004a), but the main issues were a very slow
reaction kinetics and an apparent production in very small quan-
tities. The problems were somewhat overcome herein with the
application of heat in a biphasic solvent (H2O/ethyl ether) system.
The heating of proanthocyanidins in an acidic medium induced
the production of a C4 carbocation of catechin easily attacked by
a nucleophile (vide Fig. 1), which in the present case is one of
the carbons (C6 or C8) with negative charge density (dÀ) of the
phloroglucinol ring A of malvidin-3-glucoside yielding the Cat–
Mv3glc dimeric adduct.
2.6.1. HPLC–DAD
The HPLC used was an ELITE LaChrom fitted with a L-2130 qua-
ternary pump, a L-2200 autosampler and a L-2455 DAD. The sta-
tionary phase was a MERK RP – C18 column 250 mm  4 mm i.d.
(5 lm). The injection volume was 20 ll and the mobile phase
was composed by two solvents: solvent A, a 10% (v/v) aqueous for-
mic acid solution, solvent B, formic acid:acetonitrile:water (1:3:6)
(v/v). The elution conditions were: flow rate set at 1 ml minÀ1
,
starting with a linear gradient from 20% to 66% of B in 50 min, a
second linear gradient from 66% to 100% of B for 10 min followed
by an isocratic elution of 100% of B for another 10 min. The column
was washed with 100% of acetonitrile for a 7 min period proceeded
by a re-equilibration step of 15 min of isocratic 20% of B.
The UV–vis absorbance spectra were recorded from 250 to
700 nm and the peaks recorded at 280 and 520 nm.
The synthesis was carried out in a biphasic system in order to
prevent the proanthocyanin regeneration resulting from the nucle-
ophilic attack of the phloroglucinol ring of catechin to the carbo-
cation that competes with Mv3glc (Vidal et al., 2002), and thus
enhance the reaction yield which was 12% (checked by HPLC–
DAD). The organic phase – ethyl ether – grasped the epicatechin
molecules released by the procyanidin acid decomposition away
from the aqueous phase, where the reaction between the carbocat-
ion and Mv3glc occurs.
The addition of small molar ratio quantities of procyanidins en-
sured that the concentration of Mv3glc is much higher than the
catechin unit released, improving the synthesis of Cat–Mv3glc.
The purification steps were modified given the inexistence of
conditions to perform preparative countercurrent chromatogra-
phy. The purification process proposed herein also allows an easy
scale-up.
2.6.2. HPLC–ESI-MS
The chromatographic mass spectra analyses were also per-
formed in a Finnigan Surveyor Plus HPLC system fitted with a
PDA Plus Detector, an Autosampler Plus and a LC quaternary pump
plus. The HPLC system was coupled to a mass detector Finnigan
LCQ Deca XP Plus equipped with an ESI source and an ion trap
quadrupole.
The chromatographic conditions used were: a stationary phase
composed by a Hypersil Gold 150 Â 4.6 mm i.d. (5
lm) column
(Thermofinnigan, CA). Injection volume was 25 L. Mobile phase
l
was composed by solvent A, 1% (v/v) of formic acid and solvent
B, 1% (v/v) of formic acid and 30% (v/v) of acetonitrile in water.
The flow rate was set at 0.5 ml minÀ1. The chromatographic run
started with a linear gradient from 20% to 85% of solvent B in
70 min followed by another linear gradient until 100% of solvent
B for 5 min. At this point a cleaning step of 100% of solvent B for
10 min was performed and followed by 15 min re-equilibration
time with the initial conditions.
The mass spectrometric conditions were: source and capillary
voltages of 5 kV and 4 V, respectively. Capillary temperature of
325 °C. Nitrogen was used at flow rates of 90 and 25 as sheath
and sweep gas, respectively. The mass spectra were recorded be-
tween 250 and 2000 m/z. A zoom scan of the most intense ion with
25 V collision energy for MS2 and 30 V collision energy for MS3
were also registered. For wine samples, LC–MS analyses were also
performed in Selected Ion Monitoring mode (SIM) for the signals of
m/z = 1097 and 794 followed by MS2 and MS3 fragmentations with
the conditions described as above.
To investigate the purity and unequivocally identify the syn-
thesised adduct, an NMR analysis was carried out. The results were
in agreement with the ones reported in the literature (Salas et al.,
2004b).
3.2. Hemisynthesis of Cat–Mv3glc derived compounds
3.2.1. Catechin–methylmethine–catechin-(4,8)–malvidin-3-O-
glucoside
It is well known that flavanols and anthocyanins, in the pres-
ence of acetaldehyde, condense with each other through meth-
ylmethine (Mm) bridges (Es-Safi, Cheynier, & Moutounet, 2002;
Es-Safi, Fulcrand, Cheynier, & Moutounet, 1999; Escribano-Bailon,
Alvarez-Garcia, Rivas-Gonzalo, Heredia, & Santos-Buelga, 2001;
Saucier, Little, & Glories, 1997). Bearing this, it would be expected
at least four positional isomers: Cat-8–Mm and Cat-6–Mm bridged
radicals condensed into the C8- and C6-A ring (catechin moiety) of
the Cat–Mv3glc adduct (Fig. 2). Furthermore, there is also the pos-
sibility of occurring diastereoisomers differing in the configuration
(R/S) of the asymmetric carbon of the methylmethine bridge
(Escribano-Bailon et al., 2001). Therefore, the number of com-
pounds expected in this reaction was at least eight isomers.
2.6.3. Semi-preparative HPLC
The semi-preparative HPLC system used was composed by a
Knauer K-1001 quaternary pump, a rheodyne 7125 manual injec-
tor with a 200
The stationary phase was a Thermofinnigan Hypersil Gold col-
m).
ll loop and an UV SFD S3210 detector.
umn 250 Â 10 mm i.d. (5
l
The mobile phase was composed by solvent A, 10% (v/v) of for-
mic acid, and solvent B, aqueous solution of 10% (v/v) formic acid

1132
F. Nave et al. / Food Chemistry 121 (2010) 1129–1138
The chromatogram of the reaction mixture is shown in Fig. 3.
Several peaks with retention times ranging between 30 and
60 min showed mass/charge ratios (m/z = 1097) similar to the
compound catechin–Mm–catechin-(4,8)–malvidin-3-O-glucoside.
OH
OH
OH
OH
HO
O
OH
OH
OH
HO
HO
O
+
O
OH
O
OH
H3O+, 85 ºC
OH
HO
+
OH
OH
OH
OH
OH
C4 carbocation
epicatechin
OH
Procyanidin dimer (B4)
Malvidin-3-glucoside
OH
OH
HO
O
OMe
OH
O+
OH
OH
HO
OMe
O
Glc
Cat-Mv3glc
OH
Fig. 1. Proposed reaction mechanism between procyanidin dimers and malvidin-3-glucoside under acidic conditions.
Fig. 2. Proposed reaction mechanism between catechin, acetaldehyde and Cat–Mv3glc. The structures of the four positional isomers are showed.

F. Nave et al. / Food Chemistry 121 (2010) 1129–1138
1133
m/z 781
50000
Peak D
200000
280
Cat-Mv3glc
532
25000
280
532
80000
346
Peak C
50000
280
0
534
250 300 350 400 450 500 550 600 650 700 750
wavelength (nm)
40000
0
150000
100000
352
25000
250 300 350 400 450 500 550 600 650 700 750
wavelength (nm)
356
0
250 300 350 400 450 500 550 600 650 700 750
wavelength (nm)
Cat-Mm-Cat-Mv3glc isomers
m/z 1097
D
C
B
50000
A
0
0
10
20
30
40
50
60
70
Time (min)
Fig. 3. Chromatogram (520 nm) of the reaction between Cat–Mv3glc and catechin in the presence of acetaldehyde. The main peaks with a mass/charge ratio of 1097 are
noted with capital letters A–D. Inset: Cat–Mv3glc and peaks C and D UV–vis spectra. Every noted peak present similar spectra.
Only the four major peaks noted by A–D (Fig. 3) presented consis-
tent fragmentations. Fig. 3 only shows the UV-vis spectra of two of
the noted peaks, since they are similar.
The fragmentation patterns are consistent with the structure
proposed (see Table 1). Peaks C and D are the most intense and
may correspond to the two diastereoisomers R and S resulting from
the condensation involving the C8 carbon of the phloroglucinol ring
of both catechin moieties through the methylmethine bridges by
similarity with the catechin–methylmethine dimers described in
the literature (Drinkine, Glories, & Saucier, 2005).
The two most common signals in the fragmentation pattern of
the molecular ion were m/z = 935 and 807 which correspond to
the loss of the glucose moiety ([MÀ162]+) and the loss of both
the catechin unit with subsequent rearrangement of the ethyl to
ethylene ([MÀ289À1]+), respectively (Fig. 4). The rearrangement
of the ethyl substituent was previously described by Saucier
et al. (1997).
The signal at m/z = 783 arises from the loss of a glucose moiety
followed by the loss of the B ring, by retro Diels–Alder (RDA)
decomposition in the catechin moiety of the adduct
([MÀ162À152]+).
The signal at m/z = 645 arise from the loss of catechin unit and
subsequent rearrangement of the ethyl substituent to ethylene fol-
lowed by the loss of glucose ([MÀ162À289À1]+) or vice versa
([MÀ289À1À162]+).
The mass fragment at m/z = 619, already described (Salas et al.,
2004a), resulted from the loss of catechin–methylmethine substi-
tuent and glucose moiety.
The signal m/z = 493, also previously described (Salas et al.,
2004a), arises from the loss of the A ring, by RDA decomposition
from
both
the
aglycone
fragments
of
Cat-(4,8)–Mv
([MÀ317À162À126]+) and catechin–Mm–catechin-(4,8)–malvidin
([MÀ162À442]+). Peak B displayed some unidentified signals (m/
z = 1042, 875 and 693).
Table 1
Spectral characteristics and fragmentation patterns of the peaks A–H from the
chromatograms (Figs. 4 and 7) of the reaction between catechin, acetaldehyde and
Cat–Mv3glc adduct. The most intense peaks in MS2, MS3 and the UV–vis maxima are
underlined.
As seen in Table 1, all compounds presented wavelength max-
ima at 280 and 532 nm, a shoulder in the visible band at 472 nm
coherent with the glycosilation at carbon C3 (Santos-Buelga, Gar-
Peak UV–vis kmax
m/z
m/z in MS2
m/z in MS3
cía-Viguera,
& Tomás-Barberán, 2003) and a small band at
A
1097
783, 645, 619,
493
280, 346, 472,
532
935, 783, 645, 619,
493
346 nm. Altogether, the values are consistent with the presence
of a Cat–Mv3glc moiety in the molecule (Salas et al., 2004a).
B
1097 1042, 875, 807, 693,
645
–
280, 346, 472,
532
C
1097
–
–
280, 356, 534
935, 807, 619, 535,
493
3.2.2. Methylmethine-(catechin-(4,8)–malvidin-3-O-glucoside)2
This double charged pigment was synthesised by reaction be-
tween the Cat–Mv3glc adducts and acetaldehyde. In this reaction,
was also expected the formation of some positional and diastereo-
meric isomers.
Four peaks with signal at m/z = 794 noted with E–H and other
minor peaks (not shown) were identified in the chromatogram of
the reaction (Fig. 5).
D
1097
280, 346, 472,
532
934, 806
E
794
280, 352, 468,
532
713, 632
645, 637, 632,
619
F
794
794
794
283, 472, 533
283,469, 532
280, 468, 533
713, 623
–
645, 632, 618
–
G
H
713, 632
645, 632, 619,
556
These peaks showed both similar UV–vis and mass spectra (Ta-
ble 1 and Fig. 5). There were a few negligible shifts in the wave-

1134
F. Nave et al. / Food Chemistry 121 (2010) 1129–1138
Fig. 4. Fragmentation pattern of the molecule Cat-(8,8)–methylmethine–Cat-(4,8)–Mv-3-Glc in the positive ion mode.
m/z 781
300000
280
30000
15000
Peak E
Cat-Mv3glc
280
532
533
80000
40000
468
352
352
0
200000
100000
0
250 300 350 400
500 550 600 650 700 750
wavelength (nm)
450
0
250 300 350 400 450 500 550 600 650 700
wavelength (nm)
m/z 794 Mm-(Cat-Mv3Glc)2
E
H
F
G
5
10
15
20
25
30
35
40
45
50
55
60
65
70
Time (min)
Fig. 5. Chromatogram (520 nm) of the reaction between Cat–Mv3glc and acetaldehyde. The peaks of interest with a mass/charge ratio of 794 are noted with capital letters E–
H. Inset: Cat–Mv3glc and peak E UV–vis spectra. Every noted peak present similar spectra.

F. Nave et al. / Food Chemistry 121 (2010) 1129–1138
1135
length maxima kmax but the overall spectra were similar to the
spectrum of Cat–Mv3glc.
The UV–vis spectral similarity to Cat–Mv3glc was already antic-
ipated because the molecule is composed by two Cat–Mv3glc mol-
ecules bonded through a methylmethine bridge.
Although the molecule possesses a mass of 1588 Da, it origi-
nated a mass to charge signal ratio of 794 since it presented two
charges (at each malvidin moiety). This was confirmed by a zoom
scan which showed isotopic m/z signals with a 0.5 Da difference
(Fig. 6).
794.02
80000
40000
0
794.49
794.97
795.52
795.97
790
791
792
793
794
795
796
797
798
799
m/z
Fig. 6. Zoom scan of the peak E, with the m/z ratio of 794.
OH
H
O
Glc
OMe
O+
O
H
O
H
O
H
O
H
OMe
-324 (2 x glucose)
O
H
O
OH
CH
O
H
O
H
H
3
O
H
O
H
-162 (glucose)
OH
O
O
O
H
H
O
H
OMe
OMe
O+
O
O
H
O
H
O
H
m/z 794
O
H
OMe
O+
O
H
O
H
H
m/z 632
O+
O
H
O
H
HHO
OMe
O
H
OMe
O
H
O
H
O
OH
OMe
O
Glc
O
H
OH
C
H
O
H
O
OH
3
O
H
O
H
O
C
H
O
O
H
H
3
O
H
OMe
O
O
O
H
H
O
H
OMe
O
H
O
H
O
H
O+
O
H
m/z 713
O
H
m/z 645
-943
(Cat-MvGlc
+glucose)
OMe
O
H
O+
OH
O
H
C
H
2
OMe
O
H
O
H
O
H
O
Glc
O
O
H
O
H
-969 (Mm +
Cat-MvGlc
+glucose)
OMe
O
H
O
H
-152
(RDA B
ring)
O
H
O
H
O+
-152 (RDA
B ring)
- 467 (RDA
A ring +
Mv)
OH
OMe
OH
-26 (Mm)
O
H
O
O
H
HO
O
H
OMe
O
H
O
H
OH
O
H
OH
O+
OMe
OH
O+
O
H
OMe
O+
OH
O
O
H
O
H
H
OMe
O
H
O
H
O
H
O
H
OMe
OH
OMe
m/z 619
OH
O
OH
O
OH
C
H
3
C
H
O
H
3
O
H
O
O
O
H
H
O
O
O
H
H
OMe
Glc
OMe
O
H
O
H
m/z 637
O
H
O+
O
H
O+
O
H
OMe
O
m/z 556H
OMe
O
O
H
O
H
O
H
Fig. 7. Main fragmentation pattern of the molecule Mm–(Cat-(4,8)–Mv3Glc)2 in positive mode.

1136
F. Nave et al. / Food Chemistry 121 (2010) 1129–1138
The loss of one or two glucose moieties originated signals at m/
anism described). Those factors and the reagents involved contrib-
ute together to the unsuccessful attempts performed to isolate
pure individual compounds in enough amounts for their character-
isation by NMR.
z = 713 and 632, respectively. The loss of glucose moieties as the
most common fragmentation is consistent with the fragmentation
pattern of Cat–Mv3glc (Salas et al., 2003, 2004a).
The fragmentation of the m/z = 794 ion in peaks E and H also
yielded the signals at m/z = 637 and 556. Those signals may be
interpreted as the loss of one glucose moiety and one B ring in a
catechin unit as a result of a RDA fission and the loss of two glucose
moieties and one B ring through a RDA fission, respectively (Fig. 7).
In the MS3 spectra, the most intense ion at m/z = 713 originated
mostly the fragments with m/z = 632 (loss of the other glucose
moiety), 645 (loss of Cat–Mv) and 619 (loss of Mm–Cat–Mv), the
last being the most common fragment of the Cat–Mv3glc adduct
under the same fragmentation conditions (Salas et al., 2004a).
These compounds were also detected in the reaction between
catechin, Cat–Mv3glc adducts and acetaldehyde, showing a com-
petition between flavan-3-ols and Cat–Mv3glc adducts as nucleo-
philes in the addition reaction.
The yield of both reactions to obtain individual Cat–Mm–Cat–
Mv3glc and Mm–(Cat–Mv3glc)2 compounds were shown to be
poor. In fact, the chromatograms of Figs. 3 and 5 are very complex
showing many chromatographic peaks and humps corresponding
to unidentified products resulting from side reactions involving
the reagents (acetaldehyde, Cat–Mv and catechin) that are very
likely to polymerize. On the other hand, those reactions lead to
the formation of a relatively high number of isomeric products;
four isomers were tentatively identified by mass spectrometry
for each reaction (eight isomers were expected based on the mech-
3.3. Presence of the Cat–Mv3glc derivatives in wine
The compounds previously synthesised were searched in two
different kinds of red wine samples: a table red wine (2-year-
old) and a Port wine (10-year-old) (Table 2). The compound
Mm–(Cat–Mv3glc)2 was only found in Port wine, whereas Cat–
Mm–Cat–Mv3glc was only found in the 2-year-old wine sample.
3.3.1. Two-year-old red wine sample
The six aqueous wine fractions were subjected to HPLC–MS
analysis. The fraction of 20% (v/v) acidified aqueous methanol
which revealed the presence of peaks with a signal of 1097 Da
(putative correspondence to Cat–Mm–Cat–Mv3glc) was also ana-
lysed by LC–MS SIM mode for that same signal, showing eight
peaks (Fig. 8). The chromatographic retention times of these peaks
are close to the ones from the synthesised Cat–Mm–Cat–Mv. It is
possible that some of them are derivatives of epicatehin which
have the same molecular weight, and the mass fragmentation is
expected to be similar.
Peaks 1–5 presented a fragmentation pattern consistent with
the one displayed by Cat–Mm–Cat–Mv3glc previously synthesised:
a molecular ion with m/z = 1097, which originated the ionic frag-
ments with m/z signals of 935 (main peak), 807, 781, 645 and
Table 2
Fragmentation patterns of the peaks 1–10 from the 2-year-old and 10-year-old wine samples analysed by HPLC–DAD–MS in SIM mode. The most intense peaks in SIM and MS2
are underlined.
Wine sample
2-Year-old
Peak
m/z
m/z in MS2
m/z in MS3
1
2
3
4
5
6
7
8
1097, 1098
1097, 1098
1097, 1098
1097
781, 631, 619, 494
935, 781, 631, 619, 493
781, 643, 631, 619
1015, 935, 807, 783, 744, 765, 646, 631, 619, 493
935, 807, 619, 535, 493
781, 631, 619, 494
876, 781, 644, 631, 619, 494
935, 807, 781, 730, 630, 493
1016, 935, 852, 807, 783, 659, 619, 602
935, 875, 765, 519, 493
1097
645, 637, 632, 619
645, 519, 357
1097
1097
887, 729, 644, 357
519, 357
935, 917, 773, 631, 519, 357
935, 519, 357
1097, 1098
10-Year-old
9
794
794
356, 222
433, 410
875, 775, 713, 698, 644, 632, 548, 374
746, 714, 546, 631, 596, 555, 492, 438
10
100
1
8
2
3
6
7
5
4
0
30
40
50
60
70
Time (min)
Fig. 8. Mass chromatogram (SIM at m/z = 1097) of the 20% methanolic fraction from a 2-year-old Douro wine. The peaks with signal 1097 are noted with numbers from 1 to 8.

F. Nave et al. / Food Chemistry 121 (2010) 1129–1138
1137
200
100
0
9
10
35
40
45
50
55
60
Time (min)
Fig. 9. Mass chromatogram (SIM at m/z = 794) of the 20% methanolic fraction from a Port wine. The peaks with signal 794 are noted with the numbers 9 and 10.
619. The most intense signal of MS2 fragmentation, at m/z = 935
also refragmented into 807, 781, 645 and 619.
and Mm–(Cat–Mv3Glc)2 and devise a HPLC–MS method for their
detection in wine, although this method is eminently qualitative.
These results are important because they bring insights of a part
of the chemical composition of red wines that is still practically
unknown.
Peaks 6–8 presented a molecular ion with a mass/charge signal
of 1097 and a main fragmentation signal at m/z = 935 due to the
loss of a glucose moiety, but lacked the presence of the signal at
m/z = 807, 781, 645 and 619. Instead, the fragments at m/z = 357
and 519 emerged as second and third more intense peaks of the
fragmentation of the molecular ion. The signal at m/z = 357 was
also a fragment of 935. The signal at m/z = 357 (330 + 27) and
519 (492 + 27) may correspond to malvidin and malvidin-3-gluco-
side linked to a vinyl group as a result of a loss of the catechin unit
in a Cat–Mm moiety. The signal at m/z = 519 could also be a vinyl
moiety linked to the malvidin unit of the RDA fragment (loss of the
A ring) of the Cat–Mv3glc adduct. These fragments are not compat-
ible with the ones yielded from the molecule Cat–Mm–Cat–Mv3glc
even though they would be possible in the molecule Cat–Cat–Mm–
Mv3glc, the latter one being already proposed by Atanasova, Fulc-
rand, Cheynier, and Moutounet (2002) and Alcalde-Eon et al.
(2004) resulting from the classical condensation of procyanidin di-
mer and anthocyanins mediated by acetaldehyde resulting from
oxidative wine ageing.
Acknowledgements
The authors thank the financial support of FCT (Fundação para a
Ciência e Tecnologia). F. Nave is recipient of a PhD Scholarship
SFRH/BD/30294/2006. The authors also thank Dra. Zélia Azevedo
for the LC–MS analysis.
References
Abe, Y., Sawada, A., Momose, T., Sasaki, N., Kawahara, N., Kamakura, H., et al. (2008).
Structure of an anthocyanin–anthocyanin dimer molecule in anthocyanin-
producing cells of a carrot suspension culture. Tetrahedron Letters, 49(51),
7330–7333.
Alcalde-Eon, C., Escribano-Bailón, M. T., Santos-Buelga, C., & Rivas-Gonzalo, J. C.
(2004). Separation of pyranoanthocyanins from red wine by column
chromatography. Analytica Chimica Acta, 513(1), 305–318.
Alcalde-Eon, C., Escribano-Bailón, M. T., Santos-Buelga, C., & Rivas-Gonzalo, J. C.
(2006). Changes in the detailed pigment composition of red wine during
maturity and ageing: A comprehensive study. Analytica Chimica Acta, 563(1–2),
238–254.
Alcalde-Eon, C., Escribano-Bailon, M. T., Santos-Buelga, C., & Rivas-Gonzalo, J. C.
(2007). Identification of dimeric anthocyanins and new oligomeric pigments in
red wine by means of HPLC–DAD–ESI/MSn. Journal of Mass Spectrometry, 42(6),
735–748.
3.3.2. Port wine sample
The fractions obtained from Port wine were analysed by the
same procedure as the 2-year-old red wine.
Only the 20% (v/v) methanolic fraction showed peaks with sig-
nal at m/z = 794. That fraction was reanalysed by HPLC–MS in the
SIM mode set at 794. Two peaks (retention times of 42.7 and
53.2 min) were detected with that mass/charge value (Fig. 9). Both
peaks fragmented generating signals at m/z = 713 and 632 which
are consistent with the structure of Mm–(Cat–Mv3glc)2 (Table 1).
The retention time and mass data are consistent with the struc-
ture of peaks E and H in the chromatogram of Fig. 5.
Atanasova, V., Fulcrand, H., Cheynier, V.,
& Moutounet, M. (2002). Effect of
oxygenation on polyphenol changes occurring in the course of wine-making.
Analytica Chimica Acta, 458(1), 15–27.
Borkowski, T., Szymusiak, H., Gliszczynska-Swiglo, A., & Tyrakowska, B. (2005). The
effect of 3-O-[beta]-glucosylation on structural transformations of
anthocyanidins. Food Research International, 38(8–9), 1031–1037.
Brouillard, R., Chassaing, S., & Fougerousse, A. (2003). Why are grape/fresh wine
anthocyanins so simple and why is it that red wine color lasts so long?
Phytochemistry, 64(7), 1179–1186.
Drinkine, J., Glories, Y., & Saucier, C. (2005). (+)-Catechin–aldehyde condensations:
Competition between acetaldehyde and glyoxylic acid. Journal of Agricultural
and Food Chemistry, 53(19), 7552–7558.
Drinkine, J., Lopes, P., Kennedy, J. A., Teissedre, P. L., & Saucier, C. (2007a). Analysis of
ethylidene-bridged flavan-3-ols in wine. Journal of Agricultural and Food
Chemistry, 55(4), 1109–1116.
4. Conclusions
The catechin-(4,8)–malvidin-3-O-glucoside dimeric adducts,
bearing a flavan-3-ol upper unit can undergo some of the charac-
teristic reactions of the flavan-3-ols in wine forming methylme-
thine bridged compounds. This was demonstrated either by their
hemisynthesis, in hydroalcoholic solutions in the presence of acet-
aldehyde and catechin, and by detection in red wine samples.
For the first time it was possible to present the UV-vis spectra
and mass fragmentation patterns of the Cat–Mm–(Cat–Mv3Glc)
Drinkine, J., Lopes, P., Kennedy, J. A., Teissedre, P. L.,
& Saucier, C. (2007b).
Ethylidene-bridged flavan-3-ols in red wine and correlation with wine age.
Journal of Agricultural and Food Chemistry, 55(15), 6292–6299.
Duenas, M., Fulcrand, H., & Cheynier, V. (2005). Formation of anthocyanin–flavanol
adducts in model solutions. Analytica Chimica Acta, 563(1–2), 15–25.
Escribano-Bailon, T., Alvarez-Garcia, M., Rivas-Gonzalo, J. C., Heredia, F. J., & Santos-
Buelga, C. (2001). Color and stability of pigments derived from the

1138
F. Nave et al. / Food Chemistry 121 (2010) 1129–1138
acetaldehyde-mediated condensation between malvidin 3-O-glucoside and (+)-
Pissarra, J., Lourenco, S., Gonzalez-Paramas, A. M., Mateus, N., Santos-Buelga, C., &
de Freitas, V. (2004). Formation of new anthocyanin-alkyl/aryl-flavanol
pigments in model solutions. Analytica Chimica Acta, 513(1), 215–221.
Pissarra, J., Mateus, N., Rivas-Gonzalo, J., Buelga, C. S., & de Freitas, V. (2003).
Reaction between malvidin 3-glucoside and (+)-catechin in model solutions
containing different aldehydes. Journal of Food Science, 68(2), 476–481.
Remy, S., Fulcrand, H., Labarbe, B., Cheynier, V., & Moutounet, M. (2000). First
confirmation in red wine of products resulting from direct anthocyanin–tannin
reactions. Journal of the Science of Food and Agriculture, 80(6), 745–751.
Salas, E., Atanasova, V., Poncet-Legrand, C., Meudec, E., Mazauric, J. P., & Cheynier, V.
(2004a). Demonstration of the occurrence of flavanol–anthocyanin adducts in
wine and in model solutions. Analytica Chimica Acta, 513(1), 325–332.
Salas, E., Fulcrand, H., Meudec, E., & Cheynier, V. (2003). Reactions of anthocyanins
and tannins in model solutions. Journal of Agricultural and Food Chemistry,
51(27), 7951–7961.
catechin. Journal of Agricultural and Food Chemistry, 49(3), 1213–1217.
Es-Safi, N. E., Cheynier, V., & Moutounet, M. (2002). Role of aldehydic derivatives in
the condensation of phenolic compounds with emphasis on the sensorial
properties of fruit-derived foods. Journal of Agricultural and Food Chemistry,
50(20), 5571–5585.
Es-Safi, N. E., Fulcrand, H., Cheynier, V., & Moutounet, M. (1999). Studies on the
acetaldehyde-induced condensation of (À)-epicatechin and malvidin 3-O-
glucoside in
a model solution system. Journal of Agricultural and Food
Chemistry, 47(5), 2096–2102.
Es-Safi, N. E., Meudec, E., Bouchut, C., Fulcrand, H., Ducrot, P. H., Herbette, G., et al.
(2008). New compounds obtained by evolution and oxidation of malvidin 3-O-
glucoside in ethanolic medium. Journal of Agricultural and Food Chemistry,
56(12), 4584–4591.
Fossen, T., Rayyan, S.,
& Andersen, O. M. (2004). Dimeric anthocyanins from
strawberry (Fragaria ananassa) consisting of pelargonidin 3-glucoside
covalently linked to four flavan-3-ols. Phytochemistry, 65(10), 1421–1428.
Gonzalez-Manzano, S., Perez-Alonso, J. J., Salinas-Moreno, Y., Mateus, N., Silva, A. M.
S., de Freitas, V., et al. (2008). Flavanol–anthocyanin pigments in corn: NMR
characterisation and presence in different purple corn varieties. Journal of Food
Composition and Analysis, 21(7), 521–526.
Salas, E., Le Guerneve, C., Fulcrand, H., Poncet-Legrand, C., & Cheynier, W. (2004b).
Structure determination and colour properties of
flavanol–anthocyanin dimer. Tetrahedron Letters, 45(47), 8725–8729.
Santos-Buelga, C., García-Viguera, C., Tomás-Barberán, F. A. (2003). On-line
identification of flavonoids by HPLC coupled to diode array detection. In C.
Santos-Buelga G. Williamson (Eds.), Methods in polyphenol analysis
a new directly linked
&
&
Gonzalez-Paramas, A. M., da Silva, F. L., Martin-Lopez, P., Macz-Pop, G., Gonzalez-
Manzano, S., Alcalde-Eon, C., et al. (2006). Flavanol–anthocyanin condensed
pigments in plant extracts. Food Chemistry, 94(3), 428–436.
Haslam, E. (1980). In vino veritas – Oligomeric procyanidins and the aging of red
wines. Phytochemistry, 19(12), 2577–2582.
He, J., Santos-Buelga, C., Silva, A. M. S., Mateus, N., & de Freitas, V. (2006). Isolation
and structural characterization of new anthocyanin-derived yellow pigments in
aged red wines. Journal of Agricultural and Food Chemistry, 54(25), 9598–9603.
Hemingway, R. W., & McGraw, G. W. (1983). Kinetics of acid-catalyzed cleavage of
procyanidins. Journal of Wood Chemistry and Technology, 3(4), 421–435.
Lopes-da-Silva, F., Escribano-Bailon, M. T., & Santos-Buelga, C. (2007). Stability of
pelargonidin 3-glucoside in model solutions in the presence and absence of
flavanols. American Journal of Food Technology, 2(7), 602–617.
(pp. 92–127). Cambridge: The Royal Society of Chemistry.
Saucier, C., Little, D., & Glories, Y. (1997). First evidence of acetaldehyde–flavanol
condensation products in red wine. American Journal of Ecology and Viticulture,
48(3), 370–373.
Vidal, S., Cartalade, D., Souquet, J. M., Fulcrand, H., & Cheynier, V. (2002). Changes in
proanthocyanidin chain length in winelike model solutions. Journal of
Agricultural and Food Chemistry, 50(8), 2261–2266.
Vidal, S., Meudec, E., Cheynier, V., Skouroumounis, G., & Hayasaka, Y. (2004). Mass
spectrometric evidence for the existence of oligomeric anthocyanins in grape
skins. Journal of Agricultural and Food Chemistry, 52(23), 7144–7151.
Vivar-Quintana, A. M., Santos-Buelga, C., Francia-Aricha, E., & Rivas-Gonzalo, J. C.
(1999). Formation of anthocyanin-derived pigments in experimental red wines.
Food Science and Technology International, 5(4), 347–352.