Abstract: The stability of pelargonidin 3-glucoside (Pg3gluc), the main anthocyanin in strawberry, has been studied in model citrate solutions (pH 3.5, 25½C) in the absence and presence flavanols (catechin and procyanidin B3) and/or oxygen. Changes in the solutions were monitorized by HPLC-DAD/MS and UV-visible spectra. A progressive decrease in the concentration of the anthocyanin was produced according to a first-order kinetics whose rate was influenced by the composition of the medium. The availability of oxygen was the most influential factor on the stability of the anthocyanin that decreased much slower in anaerobic conditions. On the contrary, the presence of flavanols constituted a destabilizing factor for the anthocyanin. In the absence of flavanols, the anthocyanin was mostly degraded to colourless compounds, although the formation of some unknown unstable pigments with maximum absorbance in the visible region around 400-460 nm was also observed. In the presence of oxygen the breakdown of the heterocycle was found to constitute a major pathway for anthocyanin disappearance, whereas this mechanism was hardly produced in anaerobic conditions. In the presence of flavanols the disappearance of the anthocyanin was mostly due to condensation reactions, yielding both colourless products and xanthylium pigments. Conclusions about the structure of some of the products and pigments formed on the solutions were obtained based on their mass spectra and fragmentation patterns.
INTRODUCTION
The colour is one of the main attributes through the consumer assesses food
quality and that, therefore, influences food choice and acceptability. Anthocyanins
constitute a widespread family of plant pigments that are responsible for the
colour of a large number fruits and fruit-derived products. These compounds
are unstable with regard to pH and temperature variations and other factors
such as light exposure, oxidation and the presence of metals and other substances,
giving place to colour changes during processing and storage, which reduce the
commercial value and the life span of the products containing anthocyanins.
The difficulty of strawberry-derived foods (jams, juices, concentrates or purées)
to maintain their attractive red colour is particularly known. In spite of many
studies have been carried out, there is still an insufficient knowledge of the
mechanisms subjacent to the changes of colour in strawberry foodstuffs and consequently
no definitive solutions to reinforce or maintain their colour exist. Similar
changes of colour are produced during ageing and storage of red wine, which
are attributed to the lost of original grape anthocyanins and their substitution
by complex pigments through different reactions involving anthocyanins and flavanols,
without (Liao et al., 1992; Somers, 1971; Boulton, 2001; Kunsági-Máté
et al., 2006; Hermosín-Gutiérrez et al., 2005) or
with the involvement of other compounds, such as acetaldehyde (Bakker et
al., 1993; Francia-Aricha et al., 1997; Rivas-Gonzalo et al.,
1995; Timberlake and Bridle, 1976), pyruvic acid (Bakker et al., 1997;
Bakker and Timberlake, 1997; Fulcrand et al., 1998; Romero and Bakker,
1999), vinylphenols (Cameira dos Santos et al., 1996; Fulcrand et
al., 1996; Hayasaka and Asenstorfer, 2002), a-ketoglutaric acid (Benabdeljalil
et al., 2000), acetone (Hayasaka and Asenstorfer, 2002), or glyoxylic
acid (Es-Safi et al., 1999; Fulcrand et al., 1997).
| Fig. 1: | Chemical structures of Pg3gluc (a), (+)-catechin (b) and procyanidin
dimer B3 (c) |
As a result of these reactions, new flavylium-derived compounds, pyranoanthocyanins and xanthylium pigments with different chromatic features are produced. In addition to anthocyanins, the strawberry also contains flavanols and, therefore, it can be supposed that some of the previous processes may also occur in their derived products and contribute to their changes of colour. The anthocyanin composition in the strawberry is characterised by the presence of pelargonidin 3-glucoside (Pg3gluc, Fig. 1) as major pigment, constituting 77-95% of total anthocyanins (Goiffon et al., 1999; Lopes da Silva et al., 2007). Strawberry flavanols are mainly based on catechin (de Pascual-Teresa et al., 2000b; Thompson et al., 1972), being (+)-catechin the majority compound followed by the procyanidin dimer B3 (catechin-4,8-catechin, Fig. 1); these two compounds represent around 40 and 13% of the total content of flavanols of low degree of polymerisation in the strawberry, respectively (Lopes da Silva et al., 2002). The aim of the present work was to study the behaviour of Pg3gluc in model assays in the absence and presence of oxygen and flavanols (catechin and procyanidin B3), so as to conclude about the mechanisms involved in their instability.
MATERIALS AND METHODS
Standards
(+)-Catechin, (±)-taxifolin, p-hydroxybenzoic acid and 2,4,6-trihydroxybenzaldehyde
were purchased from Sigma Chemical Co. (St. Louis, MO).
Isolation of Pelargonidin 3-Glucoside
Strawberries of the Camarosa variety were homogenized in methanol
containing 0.1% HCl, maintained at -30°C overnight and further centrifuged
(4000 g, 15 min, 2°C). Water was added to the supernatant and the methanol
removed in a rotary evaporator at 30°C. The aqueous extract obtained
was purified by liquid chromatography in an open column filled with a mixed
stationary phase composed of 20% Polyclar AT and 80% silica gel G60 to remove
sugars and acids, as described by Escribano-Bailón et al. (2002)
Isolation of Pg3gluc was further carried out by semipreparative-HPLC. An RP
Nova-Pak C18 cartridge (6 μm, 60 Å, 25x100 mm) (Waters, Milford,
MA) was used. Solvents were (A) 5% acetic acid and (B) 100% HPLC grade methanol.
The elution profile was as follows: 0 min, 90% A, 10% B; 5-20 min, 85% A, 15%
B; 20-25 min, 85% A, 15% B; 25-30 min, 82% A, 18% B; 30-50 min, 65% A, 35% B.
The flow rate was 10 mL min-1. Detection was carried out at 280 nm
and the peaks were collected in a fraction collector. Fractions containing the
anthocyanin were mixed, the solvents eliminated under vacuum and the pigment
transferred to water and freeze-dried. The purity of the anthocyanin was checked
by HPLC-DAD/MS.
Preparation of Procyanidin B3
It was synthesized as described by de Freitas (1995). The method is based
on the reduction of (±)-taxifolin by sodium borohydride followed by the
condensation between the carbonium ion (C4+), formed in acidic media
and (+)-catechin. Procyanidin B3 was isolated from the reaction mixture by fractionation
on a Sephadex LH-20 column and further purified by semi-preparative HPLC as
described by de Pascual-Teresa et al. (1998). The compound was freeze-dried
and its purity checked by HPLC-DAD/MS.
Model Solutions
Five solutions were prepared in citrate buffer 0.1M at pH 3.5. Solutions
A and B contained 2 mM Pg3gluc; solutions C and D contained a mixture of Pg3gluc
(2 mM), catechin (1.5 mM) and procyanidin B3 (1.0 mM) and solution E contained
Pg3gluc (2 mM) and procyanidin B3 (1.5 mM). All the solutions were kept in a
glass vial closed with a septum and stored in dark at 25°C. Solutions
B and D were purged with N2 during 2 min before sealing the vial. Oxygen exposure
was not avoided when vials of solutions A, C and E were closed. Samples were
taken periodically over 3 months by puncturing the septum and analysed by HPLC-DAD/MS.
Solution B was followed during 6 months. The study was conducted in 2004 at
the laboratory of Food Science, University of Salamanca (Spain).
HPLC-DAD/MS Analysis
Analyses were carried out using a Hewlett-Packard 1100 chromatograph with
a quaternary pump and a Diode Array Detector (DAD) coupled to an HP Chem Station.
An Aqua C18, 5 μm (4.6x150 mm) (Phenomenex, Torrance, CA) column was used
thermostated at 30°C. The mobile phase was (A) 0.1% trifluoroacetic acid
in water and (B) HPLC-grade acetonitrile. The elution gradient established was
0 min, 90% A, 10% B; 5-20 min, 85% A, 15% B; 20-25 min, 85% A, 15% B; 25-30
min, 82% A, 18% B; 30-50 min, 65% A, 35% B. The flow rate was 0.5 mL min-1.
Detection was carried out at 280, 330, 440 and 500 nm. MS detection was also
performed using a Finnigan LCQ detector (San Jose, CA) equipped with an ESI
source and anion trap mass analyser, which were controlled by the LCQ Xcalibur
software. The mass spectrometer was connected to the HPLC system via the DAS
cell outlet. Nitrogen was used as both auxiliary and shealth gase at flow rates
of 6 and 1.2 L min-1, respectively. The capillary voltage was 4V
and the capillary temperature 195°C. Spectra were recorded in positive
ion mode between m/z 150 and 1500. The MS detector was programmed to perform
a series of three consecutive scans: a full mass scan, a zoom scan of the most
abundant ion in the first scan and an MS-MS scan of the most abundant ion, using
a normalized collision energy of 45%.
RESULTS AND DISCUSSION
Kinetics of Pg3gluc Disappearance
The changes in the composition of the solutions were followed by analysis
of HPLC-DAD/MS. In all cases a progressive decrease in the content of Pg3gluc
was produced, although at a different rate depending on the solution. The disappearance
of the anthocyanin took place according to an apparent first-order kinetics
(C = C0. e-kt), as previously observed for the degradation of Pg3gluc
in juices (Garzon and Wrolstad, 2001) and concentrates of strawberry (Garzon
and Wrolstad, 2002; Özkan et al., 2002), as well as in model solutions
of different anthocyanins (Baublis et al., 1994; Cemeroglu et al.,
1994; Özkan, 2002; Giusti and Wrolstad, 1996). The Table
1 shows the reaction rate constants (k) and the half-life time (t1/2)
of Pg3gluc in the different solutions determined using linear regression from
plot of ln[C] vs time. For this calculation only the two first months of storage
were considered, since, except for the solution B, small or no amounts of anthocyanin
were detected after that time. The rates of disappearance were of the same order
of magnitude than the ones observed by Dueñas et al. (2006) for
the disappearance of malvidin 3-glucoside (Mv3gluc) in the presence of equimolar
concentrations of epicatechin in model solutions prepared in 20 mM ammonium
acetate at the same pH value (k = 4.92x10-2 days-1, t1/2
= 14.1 days).
As it could be expected, the disappearance of the anthocyanin was much slower when oxygen was avoided. The faster degradation in the presence of oxygen was already reported either for pure anthocyanins (Poei-Langston and Wrolstad, 1981; Kallio et al., 2000) or food matrixes (Clydesdale et al., 1978). The adjustment to a first-order kinetics was worse in the solutions containing catechin and/or procyanidin B3, suggesting that other processes may be contributing to the disappearance of Pg3gluc. Indeed, the presence of flavanols in the medium constitutes a destabilizing factor for the anthocyanin. In the conditions of the assay, the effect of the flavanols even counteracts the protection offered by the anaerobic storage, reducing the half-life of the anthocyanin to values closed to the one observed in the solutions prepared without avoiding oxygen. Poei-Langston and Wrolstad (1981) also found a destabilizing effect of the flavanols in model solutions containing Pg3gluc and catechin in a molar ratio of 1:6.7, greater than the one used in our solutions. In the assays carried out by those authors, the decrease of the anthocyanin was even faster in the mixtures Pg3gluc:catechin maintained in the absence of oxygen than in the solutions containing only anthocyanin in the presence of oxygen.
The rate of disappearance of the flavanols in the solution is slower than that of the anthocyanin (Table 1) and did not fit well a first-order kinetics, as also observed by Dueñas et al. (2006). The time of half-life was greater for catechin than for procyanidin B3, which may be explained by the lability of the interflavanic linkage and easier oxidation in the case of the dimer (Oszmianski et al., 1985; de Freitas et al., 1998; Francia-Aricha et al., 1998). Furthermore, the interflavanic cleavage of the procyanidin releases free catechin, thus contributing to moderate the decrease of this latter.
Changes in the Solutions Containing Only Pg3gluc
In the solutions containing only Pg3gluc, the disappearance of the anthocyanin
is accompanied by the formation of new compounds (Fig. 2).
The Table 2 shows the characteristics (wavelength of maximum
absorption, molecular ion and MS2 fragments) of peaks a-e for which
information could be obtained.
| Table 1: | Rate of disappearance of pelargonidin 3-glucoside (Pg3gluc),
catechin (c) and procyanidin B3 in the different model solutions assayed |
| Fig. 2: | HPLC chromatogram recorded at 280 nm of solution A at days
0, 30 and 60 of assay. The scale (mAbs units) is the same in all cases |
| Table 2: | Absorption and mass spectral characteristics of the products
found in solutions A and B |
| *Fragment ions are indicated in decreasing order of relative abundance | |
Compound a showed a molecular ion at m/z 487, to which no suitable structure could be fitted. One of its MS2 fragments corresponds to the loss of glucose (m/z at 325, -162 amu), indicating that the compound directly derives from the anthocyanin. Its higher mass in relation to Pg3gluc suggests that it could be formed by the reaction of the anthocyanin with a compound of low molecular weight, which may derive fro the citric acid used in the preparation of the solutions or a product from the degradation of the anthocyanin.
Peaks b and c were observed in solution A and identified as p-hydroxybenzoic acid (p-HB) and 2,4,6-trihydroxybenzaldehyde (THB), respectively, by comparison of their UV spectra and retention time with standards. The identity of THB could also be confirmed by its mass spectrum (molecular ion at m/z 155), whereas no ionisation of p- HB was produced in the MS conditions used. These compounds would derive from the fragments corresponding to the anthocyanin rings A (THB) and B (p-HB) released after breakdown of the heterocycle, according to the mechanism proposed by Furtado et al. (1993) and Piffaut et al. (1994). The presence of the aglycone (Pelargonidin, Pg) was irregularly detected in the chromatograms during the first of month of assay (Fig. 3), suggesting that the separation of the glucose constitutes a previous step in the anthocyanin degradation. The levels of the aglycone detected were always low in agreement with its assumed instability. THB and p-HB were not stable in the solution and progressed to the formation of other unidentified structures. The amounts of p-HB accumulated were lower and its disappearance faster than those of THB, indicating a greater reactivity. In the absence of oxygen (solution B) hardly formation of THB and p- HB was detected, denoting that the anthocyanin breakdown does not represent a relevant mechanism for anthocyanin disappearance in anaerobic conditions.
Another major product found in the solutions was peak d whose formation was observed both in the presence and absence of oxygen. In fact, this was the only important peak detected in the chromatograms of solution B maintained in anaerobic conditions (Fig. 4). This compound showed a molecular ion at m/z 257 that produced no MS2 fragments in the conditions employed. No suitable structure could be assigned to this ion. The compound e mostly accumulated in the solutions when low levels of Pg3gluc existed, suggesting that it does not derive directly from the anthocyanin but an intermediate product should exist that could not be detected.
Peak e showed a molecular ion at m/z 865 twice the mass of Pg3gluc and in its
fragmentation the successive loss of two hexose residues is produced (fragments
at m/z 703 and m/z 541). These characteristics are coherent with a dimer constituted
by two anthocyanin moieties. In assays carried out with Mv3gluc other authors
(Salas et al., 2005; Vidal et al., 2004) had already observed
the formation of anthocyanin dimers, for which flaven-flavylium C-C linked or
flavan-flavylium doubly linked structures were proposed. The compound e did
not show an absorption maximum in the region of 500 nm pointing out that it
did not contain flavylium units. Thus, two possible structures, i.e., flaven-flaven
(Fig. 5a) and doubly-linked flavan-flaven (Fig.
5b) can be suggested as possible.
| Fig. 3: | HPLC chromatogram recorded at 520 nm of solution A at day
30 of assay where the presence of pelargonidin (Pg) can be seen |
| Fig. 4: | HPLC chromatogram recorded at 280 nm of solution B at day
180 of assay |
| Fig. 5: | Suggested structures for compound e (m/z at 865) |
| Fig. 6: | Visible absorption spectra in solutions A (a) and B (b) recorded
at different days of the assay |
During storage, a progressive change in the colour of the solution from bright red to red orange and yellow orange was produced (Fig. 6). The formation of turbidity was also observed in the second month of assay and further accumulation of a brown sediment. The compounds detected in the chromatograms above discussed were mostly colourless and, therefore, other substances must have been contributing to the colour when low or no levels of the anthocyanin exist in the solution. As it can be seen in Fig. 2, in addition to peaks a-e the appearance of other peaks and a hump were also observed in the end part of the chromatograms. Those peaks, some of which showed absorption in the region of 400-460 nm, were not reproducible in the successive analyses, indicating that they corresponded to transient substances, quite possibly evolving to the formation of polymers responsible for the brown precipitate. No reliable information about those compounds could be obtained by mass spectrometry and, therefore, their nature remains unknown. The greater formation of these compounds was observed when hardly Pg3gluc existed in the medium, suggesting that they rather derived from products of the degradation of the anthocyanin and/or the citric acid used in the preparation of the model solutions. The evolution of the colour in solution B was much slower than in solution A, in accordance with the increased stability of the anthocyanin due to the restriction of oxygen. The red colour was maintained during a substantial part of the storage and only in the fourth month, when the levels of Pg3gluc were already low, a change to orange started to be appreciated. The formation of a light precipitate was also observed at the end of the six months of assay. In this solution, the products responsible for the colour when the anthocyanin decreased could not be detected in the chromatograms.
Changes in the Solutions Containing Pg3gluc and Flavanols
In solutions C-E different pattern of changes was observed with regard to
A and B. The formation of four colourless compounds (P1, P2, P5 and P6) and
four pigments (P3, P4, P7 and P8) was detected (Fig. 7) about
which information (UV-visible and mass spectra) could be obtained (Table
3). The same compounds were observed in the presence and absence of oxygen,
although in the latter case the composition changes took place more slowly.
Peak P5 corresponded to the unknown product d found in solutions A and B. This was the only important compound derived from the degradation of Pg3gluc detected in solutions C-E, where no significant formation of THB and p-HB was observed, suggesting that the heterocycle breakdown is not a preferential pathway for the anthocyanin loss in the presence of flavanols. The remaining compounds detected in solutions C-E derived from different processes of condensation between the anthocyanin and the flavanols, which constituted the most important mechanisms for the anthocyanin disappearance in these solutions, either in the presence and absence of oxygen.
The compounds P1 and P2 had the same molecular ion (m/z at 577) and similar
fragmentation pattern. Their presence in the solution E that contained the dimer
B3 as the only flavanol confirmed the implication of this procyanidin in their
formation. The molecular ion would fit A-type proanthocyanidin dimers, which
might derive from the structural rearrangement of B3 following breakage of its
interflavanic bond.
| Table 3: | Absorption and mass spectral characteristics of the products
found in solutions C-E |
| *Fragment ions are indicated in decreasing order of relative abundance | |
| Fig. 8: | HPLC chromatograms recorded at 280 nm of solution C (upper
chromatogram), solution D (middle) and solution E (bottom) after 80 days
of storage. The scale (mAbs units) is the same in all cases |
| Fig. 8: | Suggested structure for peak P1 (m/z at 577). P2 would have
similar structure but C4-C6 linked |
However, this possibility was discarded, since such type of dimers should elute later than procyanidin B3 in the chromatograms using reversed-phase columns (Santos-Buelga et al., 2003). Another possibility is that they result from the direct condensation of the carbocation C4+, released after the interflavanic breakdown of procyanidin B3, over the nucleophilic positions C8 and C6 of the carbinol form of Pg3gluc to yield F-A dimers where the anthocyanin moiety is a flavene hydrated form (Fig. 8). The MS2 fragmentation the compounds would be coherent with such a type of structure. The main fragment (m/z at 425, -152 amu) would correspond from the Retro Diels-Alder decomposition (RDA) of catechin, which has been seen to take place preferentially on the upper unit (de Pascual-Teresa et al., 2000a; Friedrich et al., 2000). Further fragmentation of this ion would give rise to the fragments at m/z 407 (loss of water) and m/z 299 (-126 amu, loss of phloroglucinol after heterocyclic ring fission of the catechin unit). Minor signals at m/z 559 and m/z 409 are interpreted as the loss of water from the hydroxyl group at C3 of the flavanol moiety (-18 amu) and catechin RDA together with the loss of the hydroxyl group at C2 of the anthocyanin moiety (-152-16 amu), respectively.
The structure proposed in Fig. 8 for P1 and P2 lacks the glucose attached to the anthocyanidin moiety. The glucosylated precursor of such a structure could be expected to be stable, however, it was not detected in any of the HPLC-DAD/MS analyses made. Thus, if the proposed structure is confirmed, it might have formed from the direct reaction of the flavanol carbocation with the anthocyanin aglycone (Pg). Taking into account that the production of the aglycone constitutes a first step in the anthocyanin breakdown (Piffaut et al., 1993), its possible reaction with the flavanol might explain the low rate of products from Pg degradation (THB and p-HB) observed in the solutions containing flavanols.
Compound P6 showed a molecular ion at m/z 723 with MS2 fragments
at m/z 571 and m/z 561 corresponding to the RDA fission of catechin and the
loss of a glucose moiety respectively. These characteristics fit a dimer containing
catechin and Pg3gluc residues. This compound was not detected in the solution
E (Pg3gluc+procyanidin B3), pointing out that the monomeric catechin was the
flavanol involved in is formation. The possibility that P6 was a catechin-Pg3gluc
dimer C6’-C8 linked was discarded, since no RDA fission of catechin could
have been produced. On the other hand, the formation of such a type of dimers
would require the previous oxidation of catechin to an o-quinone (Sarni-Manchado
et al., 1997) that would not be expected to occur in the solution D (prepared
avoiding oxygen) in which P6 is a relevant product. This observation linked
to the fact that flavanol monomers are unable to produce a C4+ carbocation,
strongly suggested that catechin was located in the lower unit of the condensed
dimer. Thus, two possible structures that fit the molecular ion found for P6
can be proposed corresponding to A-F dimers in which anthocyanin and catechin
would be either C-C linked (Fig. 9a) or doubly-linked like
in A-type proanthocyanidins (Fig. 9b).
| Fig. 9: | Suggested structures for product P6 (m/z at 723) |
In the fragmentation of anthocyanin-flavanol singly-linked species in positive ion mode, the preferential loss of the glucose moiety and further catechin RDA of the released aglycone would be expected (Gonzalez-Paramas et al., 2006; Salas et al., 2003). However, the production of a fragment from the direct catechin RDA that could even predominate over the fragment corresponding to the loss of glucose was observed for A type dimers (Dueñas et al., 2006; Remy-Tanneau et al., 2003). In our case, the main fragment produced in the MS2 fragmentation of P6 corresponded to the direct catechin RDA fission (m/z at 571), whereas a minor signal was observed for the fragment derived from the loss of the glucose (m/z at 561, 10% of relative abundance), which might support an A-type nature for P6 (Fig. 9b).
A flaven-flavan dimer like the one in Fig. 9a would be expected to be the first species formed after the condensation anthocyanin-catechin. Further rearrangement of this adduct could be produced to either a coloured xanthylium structure (Jurd, 1969; Liao et al., 1992; Somers, 1971) or a colourless A-type derivative (Jurd, 1969; Jurd and Waiss, 1965). The preferential formation of this latter in anaerobic media was found by Bishop and Nagel (1984) and Remy-Tanneau et al. (2003). In our assays P6 accumulated in higher amounts in solution D that had been purged with N2 than in solution C where oxygen was not avoided and only traces of P6 were detected. This set of observations would contribute further support to the A-type nature of this compound. On the other hand, greater formation of xanthylium pigments was observed in the solution C, suggesting that in oxidative conditions a rearrangement of the A-F adducts towards xanthylium derivatives is produced.
The products P4 (m/z at 453) and P8 (m/z at 573) showed absorption spectra corresponding to xanthylium structures (Fig. 10). In previous studies of our group the formation of a compound with the same molecular ion and similar UV-visible spectrum as P4 was found in wine model assays containing catechin and Mv3gluc (Santos-Buelga et al., 1999). That pigment was suggested to derive from the glyoxylic acid-mediated condensation between catechin and phloroglucinol moieties, in accordance with the mechanism described by Fulcrand et al. (1997). In that case, the glyoxylic acid would result from the oxidation of the tartaric acid used in the preparation of the model solutions. In the current assays the presence of glyoxylic acid in the solutions would not be expected and suggest that other mechanisms are involved in the formation of P4.
Xanthylium pigments can also be formed from the direct condensation between
flavylium cations and structures containing phloroglucinol nuclei (like flavanols)
as early described by Jurd (1969). Such a mechanism was also used by Liao et
al. (1992) and Santos-Buelga et al. (1995) to explain the formation
of yellow products in wine model solutions.
| Fig. 10: | UV-visible spectra of new xanthylium pigments formed in solutions
C-E |
Quite recently Dueñas et al. (2006), in model solutions of Mv3gluc and epicatechin in ammonium formate, found the formation of a xanthylium derivative with same molecular ion (m/z at 453) and similar UV-visible spectrum as P4 and also suggested a mechanism based on the direct condensation anthocyanin-flavanol to explain its formation. The mechanism proposed by these authors would also involve the C2-C3 oxidative cleavage of the anthocyanin heterocycle to yield an intermediate malvone, as previously described by Hrazdina and Franzese (1974). The process would require an acidic pH since under neutral or alkaline conditions the separation of the anthocyanin Bring with formation of a coumarin would be expected (Hrazdina and Franzese, 1974). Based on the proposal of Dueñas et al. (2006), a possible pathway for the formation of pigments P4 and P8 in our solutions is suggested in Fig. 11.
The structures proposed for pigments P4 and P8 in Fig. 11 fit their MS2 fragmentation patterns. Regarding P4 (m/z at 453) the main fragment at m/z 301 is attributed to catechin RDA fission [M+H-152]+; the signal at m/z 257 [M+H-152-44]+ would correspond to the additional loss of the carboxyl group. The fragments at m/z 435 and m/z 283 would derive from the loss of water [M+H-18]+ and further RDA [M+H-18- 152]+, respectively. In the case of P8 (m/z at 573) the main fragment (m/z at 435) would result of the loss of the p-hydroxybenzoyl moiety (-138 amu) corresponding to anthocyanin ring-B and the ion at m/z at 453 from the alternative breakdown of the same ring (-120 amu). Further RDA of the ion at m/z 435 would yield the fragment at m/z 283. Pigments P3 (m/z at 741) and P7 (m/z at 861) would be similar to P4 and P8, but derived from procyanidin B3 instead of catechin.
The formation of xanthylium pigments according to the mechanism of Fig.
11 involves oxidative steps, which makes it difficult to justify their presence
in the solution D prepared in anaerobic conditions.
| Fig. 12: | Proposed pathway for the formation of compounds P4 and P8 |
Obviously the incorporation of small amounts of oxygen to the vials during sampling might be an explanation. On the other hand, Somers and Evans (1986) observed that similar pattern of changes were produced in the colour composition of red wines conserved in the absence and presence of oxygen and Jurd (1967) had also showed that the flavylium cation could act as an oxidant in the process of formation of new pigments derived from the condensation between flavylium salts (like the anthocyanins) and catechin. These observations might support the formation of yellow pigments in the solution D even if oxygen was not available.
In the solutions containing anthocyanins and flavanols a change of colour was
produced from bright red to brown-orange. Furthermore, the accumulation of a
light brown precipitate was also observed at the end of the assay. In the visible
spectra of the solutions a progressive decrease in the absorbance at 500 nm
and an increase in the region of 430 nm was produced (Fig. 12),
attributable to the disappearance of the anthocyanin and the progressive accumulation
of xanthylium pigments.
| Fig. 12: | Changes in the visible absorption spectra in solutions C (a)
and D (b) at days 0 (___), 24 (----) and 80 (....)
of the assay |
The colour changes observed were similar in the three solutions containing flavanols, although they were faster in solutions C and E than in solution D, in agreement with its slower decrease of the anthocyanin and accumulation of xanthylium pigments due to the restriction of oxygen.
CONCLUSION
In this study further contributions are made to the understanding of the mechanisms involved in anthocyanin instability in food. The disappearance of Pg3gluc, by far the most important anthocyanin in strawberry, was studied in the absence and presence of flavanols and/or oxygen in aqueous acidic media. As it could be expected, the decrease of the anthocyanin was much slower when oxygen was avoided; however, the presence of flavanols constituted a destabilizing factor. Indeed, it was found that the effect of the flavanols even counteracted the protection offered by the anaerobic storage. In the solutions containing only Pg3gluc and oxygen, the anthocyanin breakdown to colourless products constituted the most important pathway for disappearance. However, in anaerobic conditions or in the presence of flavanols that mechanism does not represent a relevant way for anthocyanin loss. This, in the presence of flavanols, condensation with the anthocyanin yielding both colourless products and new xanthylium pigments was found to be the most important mechanism for anthocyanin disappearance. Conclusions about the identity of the products formed in all these reactions were obtained based on their mass spectra and MS2 fragmentation patterns and the structure of some of the newly-formed compounds was tentatively established for the first time. The results obtained are expected to contribute not only to interpret the processes of anthocyanin disappearance but they might also provide the basis to define the conditions for a better preservation of the red colour in strawberry-derived products.
ACKNOWLEDGMENTS
Financial support of Junta de Castilla y Leon (Spain) through the grant ref. SA086/01 is greatly acknowledged. Author Lopes da Silva was funded by European Union and Government of Portugal through the Programme PRODEP III (2001-2003).