EVALUATION OF COLOR AND STABILITY OF ETHYL-LINKED ANTHOCYANIN-FLAVANOL PIGMENTS IN MODEL WINE SOLUTIONS USING COMBINED CHEMICAL ANALYSIS AND 3D MOLECULAR SIMULATIONS

SUMMARY Ethyl-linked anthocyanin-flavanol pigments are one of the most important condensation products formed during the red winemaking and ageing period. They have great contribution to the color characteristics and stability of aged red wines. In this study, the color characteristics and stability of ethyl-linked anthocyanin-flavanol pigments and their precursor anthocyanins were evaluated by combined spectrophotometry and 3D molecular simulations. In model wine solutions, the condensation reactions between three anthocyanins and (-)-epicatechin, mediated by acetaldehyde, were conducted to produce ethyl-linked anthocyanin-flavanol pigments. The color was assessed by the CIELab method, and the concentration changes were analyzed by HPLC-DAD. On the other hand, the stability of these pigmented compounds was also calculated by the three 3D molecular simulation methods, that is molecular mechanics, molecular dynamics, and quantum chemistry simulation. The results obtained from CIELab analysis indicated that the formation of ethyl-linked anthocyanin-flavanol pigments resulted in a decrease of L*, a*, b* and C* values, and conversely, a rising of h* value. The 3D molecular simulations revealed that the stability of anthocyanins was as follows: Mv-3 -O- glu > Pn-3 -O- glu > Cy-3 -O- glu. The cis or trans ethyl-linked anthocyanin-flavanol pigments were much more stable than their precursor anthocyanins. Among the pigments, ethyl-linked malvidin-3 -O- glucoside-flavanol was more stable than ethyl-linked cyanidin-3 -O- glucoside-flavanol and ethyl-linked peonidin-3 -O- glucoside-flavanol.


INTRODUCTION
Color is an important sensory property of red wines and undergoes several changes during ageing from red to reddish-brown hues (Mano et al., 2007;Yuan et al., 2009;Yamagishi et al., 2010;Zifkin et al., 2012). Anthocyanins, as the color expression in wine, are extracted from grape skins through maceration during the winemaking process. However, anthocyanins from grape skins in a free state are unstable and tend to copigment with other compounds such as flavanols in red wine (Somers, 1971). Boulton (2001) reported that ethyl-linked anthocyanin-flavanol pigments formed by copigmentation can account for 50%-70% of the new wine. Ethyl-linked anthocyanin-flavanol pigments are formed essentially by condensation between anthocyanin and/or flavanols directly or mediated by aldehydes in the process of wine fermentation and ageing (Fulcrand et al., 1996;He et al., 2012). The biosynthetic pathways of these new pigments have been extensively studied in the last few decades (Timberlake and Bridle, 1976;Es-Safi et al., 1999;Mateus et al., 2002;Vivar-Quintana et al., 2002;Dueñas et al., 2006;Prat-García et al., 2020). One of them is the condensation of anthocyanins and flavanols, including direct and indirect condensation reaction. The direct condensation reaction between anthocyanins (A) and flavanols (F) cause the formation of a anthocyanin-flavanol (A+-F) adduct and a flavanol-anthocyanin (F-A+) adduct (Salas et al., 2003). The indirect condensation reaction between anthocyanins and flavanols is mediated by an aldehyde linkage (Drinkine et al., 2007;Pissarra et al., 2003). The indirect condensation reaction begins with the nucleophilic addition of the flavanol to protonated acetaldehyde. The product thus formed loses a water molecule to give a new carbocation intermediate, which proceeds the nucleophilic addition by anthocyanins in the hemiketal form, to produce ethyllinked F-Et-A or ethyl-linked F-Et-F adduct (Cheynier, 2002;Fulcrand et al., 2004). Furthermore, indirect condensation as the main polymerization reaction in wine reaction occurs quickly and starts in the fermentation process (Es-Safi et al., 2002;Dueñas et al., 2006).
During fermentation and ageing, ethyl-linked anthocyanin-flavanol pigments gradually take a dominant role, giving the aged wine a brick-red tone and mellowing the taste from its original bitterness. Sun and Spranger (2005) reported that the majority of free anthocyanins transformed into ethyl-linked anthocyanin-flavanol pigments in 'Tinta Miúda' red wine after ageing in a bottle for two years. The result of UV spectrophotometric analysis showed that the absorption wavelength of the ethyl-linked anthocyanidin-flavanol pigments was slightly higher than that of anthocyanin, that is, a red shift, indicating that the ethyl-linked anthocyanin-flavanol pigments may contribute more purple or red-purple to wine color. Besides, some studies showed that solutions containing such ethyl-linked anthocyanin-flavanol pigments (ethyl-linked malvidin-3-O-glucosideflavanol pigment) have purple hues and are more stable against acid-base changes and bleaching by sulfur dioxide than anthocyanin precursor (Bakker and Timberlake, 1997;Casassa et al., 2015;Pittari et al., 2018).
Since the reaction between anthocyanins and flavanol is in a dynamic state, throughout, it is quite difficult to isolate and prepare abundant anthocyanin-flavanol pigments from wine to study their color or stability. Most of the studies on color change of anthocyaninflavanol pigments were implemented in model wine solution, and the main anthocyanins investigated were malvidin-3-O-glucoside, while peonidin-3-Oglucoside, cyanidin-3-O-glucoside and other monomeric anthocyanins, which were rarely involved (Es-Safi et al., 2000;Escribano-Bailón et al., 2001;Pissarra et al., 2003;Sun et al., 2008). Current studies revealed that anthocyanins can transform into anthocyanin-flavanol pigments to stabilize wine color during ageing ( Sun et al., 2008;Oliveira et al., 2014), however, how their color transformation and stability change is still scarce. Molecular simulation, as a powerful compound analysis tool, may allow a multidimensional comparison of the compounds stability through various energy values (Zhao et al., 2013). Therefore, the main objective of this work was to compare the color characteristics and stability of the free anthocyanins (malvidin-3-O-glucoside, peonidin-3-O-glucoside and cyanidin-3-O-glucoside) and their indirect condensation products in model wine solution by using HPLC-DAD and CIELab method coupled with the three 3D molecular simulation methods (molecular mechanics, molecular dynamics and quantum chemistry simulation).

Condensation reaction between anthocyanins and flavanols mediated by acetaldehyde
The model wine solution for the indirect condensation reactions between anthocyanins and flavanols was prepared with 12% ethanol and 5 g/L of L-tartaric acid in water and adjusted to pH 3.2 with 1 mol/L HCl as previously reported (Sun et al., 2008). In the presence of acetaldehyde, the reactions between anthocyanins and (-)-epicatechin were carried out in the model wine solution under the optimal conditions (molar ratio of anthocyanin/epicatechin/acetaldehyde: 1/6/10; reaction temperature 35℃) based on the previous reported study (Li et al., 2018).

HPLC-DAD chromatographic conditions
The anthocyanins, (-)-epicatechin and their ethyllinked anthocyanin-flavanol pigments were analyzed based on Waters e2695 Alliance HPLC including a quaternary pump, a controller, an autosampler , a 2998 PDA detector working in the range of 200-800 nm and a data processing work station. The optimized chromatographic conditions were demonstrated in the previously published research works (Li et al., 2018). Briefly, the column was Innoval C18 (5 µm, 4.6 × 250 mm) maintaining at 30℃. The flow rate was fixed at 0.7 mL/min. Solvent A (water:formic acid; 98:2, v/v) and solvent B (water:acetonitrile:formic acid; 68:30:2, v/v) were used with the following elution gradient: 0 min, 18% B; 42-48 min, 47% B; 78-110 min, 100% B; and flush and balance of the column for 10 min. The detection wavelength was 525 nm for the detection of anthocyanins and their derivatives, and 280 nm for all polyphenols.

ChemBioDraw
Ultra 14.0 (CambridgeSoft Corporation, USA) was used to create the twodimensional structure of the compounds. The twodimensional constructions were converted into threedimensional structures using Sybyl-6.91 (Trepos Corporation, USA). The maximum iteration was set to 10000, the force field charge was Gasteiger-Huckel, and small molecule mechanics optimization and simulation were carried out and the required parameters were obtained.

Molecular dynamics method
Molecular dynamics simulation was based on the optimal conformation. Simulated Annealing could find the global minimum energy. Polak-Ribiere algorithm was used to carry out the geometric optimization of single point calculation and obtain the optimized configuration. Simulated Annealing was selected for molecular dynamics simulation optimization based on the optimized configuration in this experiment to obtain the lowest global energy configuration and corresponding energy values. The reference values were as follows: time increment for dynamics computations was 0.5 fs; coupling time for temperature regulation was 2.0 fs; initial temperature for heating was 426.85℃ ; time to equilibrate at initial temperature was 1,000 fs; target temperature for annealing was -73.15℃; time to spend annealing was 1,000 fs. These steps were used to obtain a stable, balanced conformation with the lowest energy.

Quantum chemical method
The lowest energy and optimal three-dimensional conformation were obtained through molecular mechanics and molecular dynamics methods with HyperChem 8.0 software (Hypercube, USA). Single point calculations and structure optimization were carried out separately using semi-empirical PM3 quantum chemical methods.

Molecular simulations in aqueous solutions
A boundary condition of 20 × 20 × 20 was set to add 265 water molecules to simulate the spatial conformation and energy difference of anthocyanins and ethyl-linked anthocyanin-flavanol pigments in an aqueous solution. The minimum distance between the solvent and solute atom was set as 2.3 for all models.

Statistical analysis
All statistical analyses were performed using SPSS (Version 22.0, Chicago, IL, USA). The results were performed in three independent experiments for each sample. The results were analyzed by one-way analysis of variance (ANOVA) followed by Duncan's multiple range test.

Formation and identification of ethyl-linked anthocyanin-flavanol pigments in the indirect condensation reactions
Mv-3-O-glu, Pn-3-O-glu and Cy-3-O-glu were respectively used as a reactant to react with (-)epicatechin to conduct the condensation reaction in model wine solution. The condensation reaction was monitored continuously for 224 hours. After the reaction started, samples were taken every two hours, filtered through a 0.22 μm filter membrane and analyzed under HPLC-DAD chromatographic conditions. At the beginning of the reaction, precursor free anthocyanins and (-)-epicatechin were detected by HPLC analysis as shown in Figure 1(A, D, G). The condensation reaction was carried out after 2 h, and the ethyl-linked anthocyanin-flavanol pigments were gradually formed as presented in Figure 1 (B, E, H). After the reaction proceeded to 224 h, there were not only predominantly ethyl-linked anthocyanin-flavanol pigments in the presence of the model wine solution system, but also formed of by-products, and presented in Figure 1 (C, F, I).
Formed ethyl-linked anthocyanin-flavanol pigments were identified based on primary and secondary fragment ions using HPLC-QTOF-MS as described in the previous study (Li et al., 2018). The results showed that peaks 3 and 4, peaks 6 and 7, peaks 9 and 10 as isomers were consisted of (-)-epicatechin and anthocyanin linked by ethyl group, with the same molecular formula, molecular weight and cleavage pattern, but with different configuration and conformation. The absolute configuration and conformation of ethyl-linked anthocyanin-flavanol pigments were determined by ECD analysis. The results indicated that peak 3 was identified as Sconfiguration ethyl-linked cyanidin-3-O-glucosideepicatechin (Cy-ethyl-EC), peak 6 was identified as Sconfiguration ethyl-linked malvidin-3-O-glucosideepicatechin (Mv-ethyl-EC) and peak 9 was identified as S-configuration ethyl-linked peonidin-3-Oglucoside-epicatechin (Pn-ethyl-EC) in Figure 1. Peak 4 was named as R-configuration Cy-ethyl-EC, peak 7 was named as R-configuration Mv-ethyl-EC, and peak 10 was named as R-configuration Pn-ethyl-EC in Figure 1.

Content changes of anthocyanins and ethyl-linked anthocyanin-flavanol pigments during the indirect condensation reactions
As shown in Figure 2, the contents of Cy-3-O-glu, Mv-3-O-glu and Pn-3-O-glu, as well as their associated ethyl-linked anthocyanin-flavonol pigments produced by interaction with (-)-epicatechin, were dynamically monitored by HPLC. Reactions of Mv-3-O-glu, Pn-3-O-glu and Cy-3-O-glu with (-)-epicatechin showed the same pattern of variation. The anthocyanin content decreased significantly from 0 to 40 h. The contents of Cy-3-O-glu, Mv-3-O-glu and Pn-3-O-glu were reduced from 0.817 mg/mL to 0.017 mg/mL, 0.82 mg/mL to 0.004 mg/mL, 0.78 mg/mL to 0.0076 mg/mL, respectively. From 40 to 340 h, the decrease of the anthocyanin content tended to level off until below the limit of detection, indicating that they were basically completely transformed into ethyl-linked anthocyanin-flavanol pigments. Burtch et al. (2017) reported that wines containing high concentrations of diglucoside anthocyanins will form less polymeric pigment than wines containing primarily monoglucoside anthocyanins. In that study, monoglucoside anthocyanins were also converted to polymeric pigment approximately 7.5 times more quickly than diglucoside anthocyanins. Furthermore, monoglucoside anthocyanins reached nondetectable levels within 14 days (336 h .0178, 0.0306, and 0.0394 mg/mL, respectively. In this study, the pH 3.2 and normal oxygen level was used to simulate the matrix of real wine, and the content change of reactants and products was mainly investigated by reaction time. However, for a polymerization reaction, the presence of oxygen and lower pH values can promote it, since the formation of acetaldehyde and its protonated form is favored under such conditions (Rivas-Gonzalo et al., 1995;Heredia et al., 1998). Therefore, the influence of these factors on changes of ethyl-linked anthocyanin-flavanol pigment content requires further investigation.

Reaction system of Cy-3-O-glu and epicatechin in the presence of acetaldehyde
For the precursor Cy-3-O-glu solutions, L* value was 42.39, a* value was 6.41, b* value was 6.24, C* value was 8.95, h* value was 52.3, and ∆E* value was 5.67. Figure 3 (A) depicted the change in color characteristics during the reaction between Cy-3-Oglu and (-)-epicatechin. L* value decreased from 42.39 to 31.48 in the 0-3 days, remained stable during 10 days, and then significantly decreased and remained stable; a*, b* and C* values decreased significantly from 0-3 days, barely fluctuated from 3-10 days, and significantly increased after 11 days; the h* value showed a significant rise trend from 52.3 to 273.09 in 0-3 days, which was the same change trend as that of L* value; the △E* value decreased significantly in 0-3 days, then increased slowly in 3-9 days, followed by sustained fluctuations. As shown in Figure 3, the twodimensional color parameters and color comparison between the initial and final reaction points for Cy-3-O-glu indicate a variation of an initial orange tone to a final brick red tone.

Reaction system of Mv-3-O-glu and epicatechin in the presence of acetaldehyde
For precursor Mv-3-O-glu solutions, L* value was 43.62, a* value was 4.79, b* value was 5.26, C* value was 7.12, h* value was 51.75, and ∆E* value was 4.2. Figure 3 (B) depicted the change in color characteristics during the reaction between Mv-3-Oglu and (-)-epicatechin. L* value decreased from 43.62 to 31.64 in the 0-3 days, remained stable during 10 days, and then significantly decreased and remained stable; a* and b* values significantly decreased in the 0-3 days, and showed no fluctuation in the 3-10 days, followed by a great increase after 11 days; C* value significantly decreased in the 0-1 day, remained stable in the 1-10 days, and decreased and remained stable after 11 days; the h* value showed a significant rise trend from 51.75 to 272.9 in the 0-3 days, which was the same as that of L* value; the ∆E* value decreased significantly in 0-3 days, then slowly increased in 3-9 days, followed by continuous fluctuations. As shown in Figure 3, the two-dimensional color parameters and color comparison between the initial and final reaction points for Mv-3-O-glu indicate a variation of an initial red tone to a purple tone during the reaction.

Reaction system of Pn-3-O-glu and epicatechin in the presence of acetaldehyde
For precursor Pn-3-O-glu solutions, L* value was 41.78, a* value was 6.42, b* value was 6.08, C* value was 8.85, h* value was 52.4, and ∆E* value was 6.24. Figure 3 (C) depicted the change in color characteristics during the reaction between Pn-3-O-glu and (-)-epicatechin. The results showed that the L* value decreased from 41.78 to 31.22 in the 0-3 days, and the change was stable in 3-10 days, and then a significant decrease occurred after 11 days; the a* and b* values showed a significant decrease in the 0-3 days, a more stable change with almost no fluctuation in the 3-10 days, and remained stable after a significant increase in the following 11 days; the C* value showed a significant decrease in 0-1 day, remained stable in 1-10 days, and showed an increase and remained stable after 11 days; the h* value showed a significant rise trend from 52.4 to 273.13 in the 0-3 days, which was the same change trend as that of L* value; the △E* value significantly decreased from 6.24 to 2.23 in the 0-1 day, and remained stable in the 2-13 days. As shown in Figure 3, the two-dimensional color parameters and color comparison between the initial and final reaction points for Pn-3-O-glu indicate a variation of the initial orange tone to a brick red tone during the reaction. . Therefore, it can be concluded that hydroxyl groups contributed more to both the decrease of L* values and the increase of a*, b*, C*, h* and ΔE* values than the methoxyl groups. This result is consistent with those of the study of Heredia et al. (1998), in which it was concluded that the absorbance decreased when hydroxyl groups were substituted by methoxyl groups, according to the comparison of two (Cy-3-O-glu > Pn-3-O-glu) B-ring substituted anthocyanins. Therefore, it is suggested that to a large extent, the color of the three common anthocyanin depended on the number, position and type of substituents on B-ring.
Based on the results above mentioned, significant differences in color characteristics were observed among the different reaction systems. In the process of reaction between anthocyanins (Cy-3-O-glu, Mv-3-Oglu and Pn-3-O-glu) and (-)-epicatechin, during 0-3 days, L* values of Cy-3-O-glu, Mv-3-O-glu and Pn-3-O-glu gradually decreased, remained stable for a while and decreased significantly after 11 days. C* value presented a similar variation trend to that of a* and b* for Cy-3-O-glu, Mv-3-O-glu and Pn-3-O-glu. In short, it mainly showed that a*, b* and C* significantly decreased in the 0-3 days and remained stable in the 3-10 days, then rise and remain stable after 11 days. h* values of Cy-3-O-glu, Mv-3-O-glu and Pn-3-O-glu presented an opposite change compared with a*, b* and C*, which significantly increased in the 0-3 days and remained stable in the 3-10 days with no significant fluctuations in data. Escribano-Bailón et al. (2001) found that in the acetaldehyde-mediated condensation between Mv-3-O-glu and (+)-catechin, a decrease occurred in the L* values and h* values. The variation tendency of h* values was opposite to the results of the present study. It was possible that cistrans isomers of flavan-3-ol have an effect on h* value. The decrease of ∆E* value was fastest for Pn-3-O-glu, followed by Cy-3-O-glu and Mv-3-O-glu in the 0-3 days. This behavior was likely due to the slightly higher reaction rate of Pn-3-O-glu with (-)-epicatechin than that of the other two anthocyanins. To determine if the changes in chromatic parameters were noticeable, the color differences (∆E* values) among three reaction system were calculated. Previous research has shown that even an untrained human eye can distinguish between two colors with a ∆E* value of 3.0 units (Pissarra et al., 2003). The obtained results revealed that the ∆E* value was consistently higher than 4 for each reaction system, indicating a noticeable color change above the perceptible threshold and making it always detectable (Spagna et al., 1996). The noticeable change in color before and after the reaction in model wine solution was clearly presented in Figure 3, providing strong evidence for the reliability of the experimental results. The formation of ethyl-linked anthocyanin-flavanol pigments promoted a decrease of L*, a*, b* and C* values and, conversely, a rising of h* value. The results were consistent with those of Dufour and Sauvaitre (2000), according to whom these ethyllinked anthocyanin-flavanol pigments are more stable than monomer anthocyanins and contribute to stabilizing wine color. (2) final point as brick-red hue (Chromatic circle: +a* redness to greenness -a*; +b* yellowness to blueness -b*).

Stability of three-dimensional structure model of anthocyanins and ethyl-linked anthocyanin-flavanol pigments by molecular mechanics method
2D structural models of anthocyanins and ethyl-linked anthocyanin-flavanol pigments were presented in Figure 4. The 2D structure was transformed into a 3D structure using Sybyl-6.91. Following molecular mechanics optimization, the simulated molecular structure was depicted in Figure 5, and the energy values are shown in Table Ⅰ. The bond energy (bond stretching energy, angle bending energy, torsional energy and out of plane bending energy) represents the strength of the bond. It was important to consider bond energy value when estimating the stability of the compounds (Liu et al., 2017). It was clear that the bond energy of the three anthocyanins was Mv-3-O-glu > Pn-3-O-glu > Cy-3-O-glu, therefore, it was speculated that the 3' and 5' methoxyl groups on anthocyanin B ring are more difficult to fracture than hydroxyl groups with the same position on anthocyanin B ring. The bond energy of either cis or trans ethyl-linked anthocyanin-flavanol pigments was higher than that of the corresponding anthocyanins. Hence, it indicated that the ethyl-linked anthocyanin-flavanol pigments structure was more stable than the anthocyanin one. The high contribution of van der Waals energy leads to a general contraction and higher stability of compounds (Lan et al., 2013). Regardless of Mv, Pn, and Cy, ethyl-linked anthocyanin-flavanol pigments have high van der Waals forces compared to monomer anthocyanins, indicating that ethyl-linked anthocyanin-flavanol pigments had good stability. In terms of total potential energy, the compounds were sorted as follows:

Table Ⅰ
Energy values obtained for the molecular modeling of anthocyanins and ethyl-linked anthocyanin-flavanol pigments with molecular mechanics

Stability of three-dimensional structure model of anthocyanins and ethyl-linked anthocyanin-flavanol pigments by molecular dynamics method
The 3D conformation of the anthocyanin and ethyllinked anthocyanin-flavanol pigment after optimization by simulated annealing is shown in Figure 6, and the lowest energy values are presented in Table Ⅱ. In molecular dynamics, there is a close correlation between the energy value of a compound and its activity level, with higher energy values corresponding to more active compounds (Ai et al., 2022).The energy of Mv-3-O-glu, Pn-3-O-glu, and Cy-3-O-glu were 99.443, 98.911 and 85.362 kcals/mol, respectively. Therefore, the energy values of the three anthocyanins were Mv-3-O-glu > Pn-3-Oglu > Cy-3-O-glu. The energy of six ethyl-linked anthocyanin-flavanol pigments was higher than those of monomer anthocyanins. Among them, the order of energy was: Mv-ethyl-EC (R) (158.767 kcal/mol) >

Stability of three-dimensional structure model of anthocyanins and ethyl-linked anthocyanin-flavanol pigments by quantum chemical method
The computation of anthocyanins and ethyl-linked anthocyanin-flavanol pigments models was carried out by geometric optimization using the molecular mechanic method in the MM + , AMBER and OPLS force fields of HyperChem 8.0. A parameterized model 3 (PM3) semi-empirical quantum chemistry method was used to calculate the heat of formation (Zhao et al., 2013). Comparing the heat of formation for each conformation after optimized 3D conformations, low values showed a positive correlation with structural stability (Zhao et al., 2013). According to Table Ⅲ, in the three force fields, no difference was found in the heat of formation for the same compound. The order of heat of formation was: Table Ⅲ Calculated properties for anthocyanins and ethyl-linked anthocyanin-flavanol pigments with different molecular mechanics methods in vacuum by the single energy of method PM3

Differences in the spatial conformation and energy of anthocyanins and ethyl-linked anthocyanin-flavanol pigments in molecular simulations of aqueous solutions
In the aforementioned quantum chemical optimization, there was no discernible variation in the heat of production under three different molecular force fields. Therefore, the MM + force field was selected as the typically force field to optimize the conformation of anthocyanins and ethyl-linked anthocyanin-flavanol pigments under simulation aqueous medium and to compare their stability. Molecular mechanics, molecular dynamics and quantum chemistry was carried out in a vacuum condition, whereas anthocyanins and ethyl-linked anthocyanin-flavanol pigments were actually presented in aqueous conditions. Therefore, it was focused on investigating the effect of anthocyanins and ethyl-linked anthocyanin-flavanol pigments on their conformation and stability under simulation aqueous medium. Figure 7 showed the optimized three-dimensional conformation of anthocyanins and their ethyl-linked anthocyanin-flavanol pigments in aqueous solutions. The solvation and non-solvation energy values were compared, as shown in Table Ⅳ. The electrostatic energy of all compounds changed from zero to a negative value, and the van der Waals energy and bond energy significantly increased, indicating that each compound was more stable in aqueous medium (Lan et al., 2013;Liu et al., 2017).  Compared to the bond energy of ethyl-linked anthocyanin-flavanol pigments and precursor anthocyanins, it was found that the bond energy of the former was higher than that of the latter, indicating that ethyl-linked anthocyanin-flavanol pigments were more stable in aqueous solution. In short, compared to precursor anthocyanins, ethyl-linked anthocyaninflavanol pigments showed an obvious stability.
In conclusion, three anthocyanins ( Pn-ethyl-EC (R)) were simulated by molecular mechanics, molecular dynamics and quantum chemistry. The results showed that these nine compounds were relatively stabled in the force field.
By comparing bond energy, the heat of formation, and other energy parameters, it was found that ethyl-linked anthocyanin-flavanol pigments were more stable than their corresponding precursor anthocyanins. In particular, ethyl-linked malvidin-3-O-glucosideflavanol had distinctive characteristics of stability. Conformational simulations of anthocyanins and ethyl-linked anthocyanin-flavanol pigments were carried out in MM + force fields to study their stability under simulation in aqueous medium. It was found that the stability of anthocyanins and ethyl-linked anthocyanin-flavanol pigments in simulation aqueous medium was enhanced compare to that a vacuum condition in MM + force field, and the ethyl-linked anthocyanin-flavanol pigments were more stable than the monomer anthocyanins in aqueous solution. It was confirmed that the ethyl-linked anthocyanin-flavanol pigments were more stable than the anthocyanins from the energy point of view.

CONCLUSIONS
In this work, using model wine solution system, HPLC-DAD and CIELab analyses were performed to monitor changes in the contents and color parameters in the process of reaction between anthocyanins (Mv-3-O-glu, Pn-3-O-glu and Cy-3-O-glu) and (-)epicatechin. The formation of ethyl-linked anthocyanin-flavanol pigments was responsible for the changes in color parameter values: decrease of L*, a*, b* and C*, and increase of h*. The color of the Cy-3-O-glu and Pn-3-O-glu reaction system shifts from the initial orange-red hue to a brick-red hue, and the color of the Mv-3-O-glu reaction system shifts from an initial red hue to violet hue. To the best of our knowledge, different 3D molecular simulation methods were used, for the first time, to calculate the stability of the three major anthocyanins in red wine (Mv-3-O-glu, Pn-3-O-glu, Cy-3-O-glu) as well as their indirect condensation products anthocyanin-flavanol pigments (Cy-ethyl-EC (S), Cy-ethyl-EC (R), Mvethyl-EC (S), Mv-ethyl-EC (R), Pn-ethyl-EC (S) and Pn-ethyl-EC (R)). It was shown that the stability of three anthocyanins was Mv-3-O-glu > Pn-3-O-glu > Cy-3-O-glu. Besides, ethyl-linked anthocyaninflavanol pigments were found to be more stable than free precursor anthocyanins, particularly, the stability of the ethyl-linked malvidin-3-O-glucoside-flavanol was remarkable. Under simulation in aqueous medium, the stability of ethyl-linked anthocyanin-flavanol pigments was higher than that of their precursor anthocyanins, and was as follows: Mv-ethyl-EC (S) > Pn-ethyl-EC (S) > Cy-ethyl-EC (R) > Mvethyl-EC (R) > Pn-ethyl-EC (R) > Cy-ethyl-EC (S). In conclusion, the present study used a combination of chemical analysis and 3D molecular simulations, confirming that the stability of the major anthocyanins in red wine can be significantly enhanced by indirect reaction with flavanols to form ethyl-linked anthocyanin-flavanol pigments during the ageing process.