Simultaneous Determination of Nonanthocyanin Phenolic Compounds in Red Wines by HPLC-DAD/ESI-MS

Materials.

Methyl gallate, ethyl gallate, quercetin-3-O-galactoside, quercetin-3-O-glucoside, kaempferol-3-O-glucoside, and (−)-epicatechin-3-O-gallate were purchased from Extrasynthèse (Genay, France). (+)-Catechin, (−)-epicatechin, myricetin, quercetin, kaempferol, tryptophol, trans-resveratrol, and gallic, trans-caffeic, trans-p-coumaric, trans-ferulic, vanillic, protocatechuic, and el-lagic acids were purchased from Sigma (St. Louis, MO). Syringic acid and tyrosol were purchased from Aldrich (Deisenhofen, Germany). Cis-resveratrol was obtained from trans-resveratrol after exposure to UV light (340 nm) for one hour. Procyanidin dimers B1, B2, B3, and B4, trimer C1 [(−)-epicatechin-(4β→8)-( −)-epicatechin-(4β→8)-( −)-epicatechin], and trimer T2 [(−)-epicatechin-(4β→8)-( −)-epi-catechin-(4β→8)-(+)-catechin] were isolated from a grape seed extract and their structure elucidated as previously described (Monagas et al. 2003).

Winemaking.

Wines from grapes of Vitis vinifera L. cv. Tempranillo, Graciano, Cabernet Sauvignon, and Merlot (vintage 2000) of the same age and cultivated in the same vineyard (Navarra, Spain) were vinified at the Viticulture and Enology Station of Navarra (EVENA), Olite, Spain, under identical conditions. A lot of 220 kg of grapes of each variety was destemmed, crushed, and collected into 200-L stainless-steel wine vats. Semi-industrial scaled fermentations were performed with a yeast inoculum of 25 g/hL (80% EVENA Saccharomyces cerevisiae strain Na33; 20% Lallemand Saccharomyces bayanus strain EC118) at a temperature up to 27°C. The cap was punched down twice a day until it remained submerged during a 14-day maceration at the same temperature. At the end of the alcoholic fermentation (sugar level <2 g/L), the wines were racked and the grape pomace was pressed using a vertical pneumatic press at 2 bars, recovering ~50 L of wine, which was combined with nonpressed wines. After one month of stabilization at −2°C, the wines were racked, filtrated with SEITZ K250 filters (2.5 to 3.0 mm) (Sert Schenk Filter System GmB, Bad Krevznach, Germany), and bottled after correcting the free SO₂ level to 30 mg/L. Two wine samples from each variety were analyzed after 1.5 months of bottling and storage at 13°C and 80 to 85% relative humidity.

Extraction of phenolic compounds.

A volume of 50 mL of wine was concentrated to 15 mL under vacuum at 30°C and extracted three times with diethyl ether (15 mL) and three times with ethyl acetate (15 mL), as described by Fernández de Simón et al. (1990). The organic phases were combined and dried with anhydrous Na₂SO₄ for 30 min. The extract was then taken to dryness under vacuum, dissolved in 2 mL of methanol/water (1:1), then filtrated (0.45 mm) and injected into the HPLC column. Recovery of phenolic compounds in diethyl ether and ethyl acetate varies with chemical structure (mean recovery value = 87%; mean SD value = 6%) (Fernández de Simón et al. 1990).

HPLC–DAD/ESI–MS.

A Hewlett-Packard series 1100 (Palo Alto, CA) chromatography system equipped with a diode array detector (DAD) and a quadruple mass spectrometer (Hewlett-Packard series 1100 MSD) with an electrospray interface was used. Separation of 5 to 10 mL of wine extract was performed on a reversed-phase Waters Nova-Pak C₁₈ (300 mm x 3.9 mm, 4 mm) (Millipore, Milford, MA) column at room temperature. A gradient consisting of solvent A (water/acetic acid, 98:2, v/v) and solvent B (water/acetonitrile/acetic acid, 78:20:2, v/v/v) was applied at a flow rate of 0.7 mL/min as follows: 0 to 80% B linear from 0 to 55 min, 80 to 90% B linear, from 55 to 57 min, 90% B isocratic from 57 to 70 min, 90 to 95% B linear from 70 to 80 min, 95 to 100% B from 80 to 90 min, followed by washing with methanol and reequilibration of the column from 90 to 120 min. The electrospray ionization (ESI) parameters were drying gas (N₂) flow and temperature, 10 L/min and 350°C, respectively; nebulizer pressure, 55 psi; and capillary voltage, 4000 V. The ESI was operated in negative mode scanning from m/z 100 to 3000 using the following fragmentation program: from m/z 0 to 200 (100 V) and from m/z 200 to 3000 (200 V). Diode array detection was performed from 220 to 380 nm. Quantification was carried out by external standard calibration curves. Hydroxybenzoic acids, stilbenes, phenolic alcohols and other related compounds, flavanols, and flavonols were quantified at 280 nm, caffeic acid and its derivatives at 340 nm, and p-coumaric acid and its derivatives at 310 nm. Caffeic and p-coumaric acid derivatives, flavonol glycosides, and stilbene glucosides were quantified by the calibration curve of their respective free forms. Monomeric and dimeric flavan-3-ols were quantified using the (−)-epicatechin calibration curve. The remaining phenolic compounds were quantified by calibration curves produced with the same compound.

Identification of phenolic compounds.

An HPLC-DAD chromatogram of the diethyl ether/ethyl acetate extract from the Tempranillo wine is shown in Figure 1⇓ (peaks cited correspond to those in Table 1⇓). Nonanthocyanin phenolic compounds were identified by peak retention time, UV spectra, and ESI-MS spectrometric data (Table 1⇓). Nonflavonoid phenolic compounds (hydroxybenzoic and hydroxycinnamic acids and their derivatives, stilbenes, and phenolic alcohols and other related compounds) and flavonoid phenolic compounds (flavanols, flavonols, and dihydroflavonols) were identified in the wines from the four different grape varieties.

• Download figure
• Open in new tab

Figure 1

HPLC-DAD chromatogram at 280 nm of nonanthocyanin phenolic compounds of diethyl ether/ethyl acetate extract from Tempranillo wine.

Hydroxybenzoic acids and their derivatives.

Gallic (1), protocatechuic (3), vanillic (20), and syringic (25) acids were the main hydroxybenzoic acids identified in the different wines. For gallic and protocatechuic acids, in addition to the molecular ion, a fragment ion [M-H-44]⁻ corresponding to the loss of the -CO₂ group (carboxylic function) was detected (Vanhoenacker et al. 2001) (Table 1⇑). Other authors (Pérez-Magariño et al. 1999) have also observed differences in the fragmentation pattern of hydr-oxybenzoic acids according to their chemical structure.

Peaks 7 and 30 with [M-H]⁻ at m/z 183 and 197 were identified as methyl gallate and ethyl gallate, respectively (Figure 2A⇓). The fragment ions at m/z 169 and 125, corresponding to gallic acid after the loss of the methyl or ethyl unit and to the subsequent loss of the -CO₂ group, respectively, were detected in very low concentrations. Peak 18 presented molecular and fragment ions (m/z 329 and 167), which coincided with the mass of an hexose derivative of vanillic acid, being either the hexose ester or the glycoside. However, the peak intensity for this compound in the HPLC-DAD was too weak to allow further characterization on the basis of UV spectral characteristics. Both ethyl gallate and hexose (glucose) esters of vanillic acid have been previously identified in Riesling wine (Baderschneider and Winterhalter 2001). Methyl gal-late has been previously reported in red grapes (Fernández de Simón et al. 1992). Despite presenting a lactone-type structure, ellagic acid (peak 40) is included in this subgroup of compounds, as it is formed by oxidative coupling of two vicinal galloylated ester groups followed by acid hydrolysis (Haslam 1998).

• Download figure
• Open in new tab

Figure 2

Comparison of ultraviolet and ESI-MS spectra: (A) methyl and ethyl gallates; (B) hexose esters of trans-p-coumaric acid.

Hydroxycinnamic acids and their derivatives.

trans-Caftaric (5), coutaric (14), and fertaric acids (17) as well as trans-caffeic (22) and p-coumaric (32) acids have been identified. For the tartaric esters, a fragment ion [M-H-132]⁻ corresponding to the free acid after the cleavage of the ester bond was observed (Table 1⇑). The fragment ion corresponding to the loss of the -CO₂ group was mainly detected for the free acids, as observed in the case of hydroxybenzoic acids and their derivatives.

Peaks 23 and 27 presented similar cinnamic-type UV spectra, molecular ion ([M-H]⁻ = m/z 325), and fragments [m/z 163 ([M-H-glucose]⁻); m/z 145 ([M-H-glucose-H₂O]⁻)] (Figure 2B⇑), coinciding with the mass of an hexose moiety linked to trans-p-coumaric acid (Baderschneider and Winterhalter 2001, Biau et al. 1996). However, because O-glycosides present a maximum absorption wavelength hypsochromically shifted with respect to that of the free acid (Baderschneider and Winterhalter 2001, Määttä et al. 2003), which was not observed for peaks 23 and 27, we tentatively propose the ester-type structure. The origin of the intermediate fragment at m/z 265 (Figure 2B⇑), corresponding to the loss of 60 amu (-C₂H₄O₂), could not be elucidated, although it has also been detected in berries for two compounds exhibiting the same UV and mass spectra characteristics as peaks 23 and 27 (Määttä et al. 2003). After considering similar descriptions in the literature (Macheix et al. 1990), we believe these compounds are either forms composed of two different hexoses or isomers composed of the same hexose esterified through different -OH positions of the molecule. Hexose (glucose) esters of trans-p-coumaric acid have been identified previously in Riesling wines (Baderschneider and Winterhalter 2001).

Another compound with phenylpropanoid structure corresponded to peak 6, which presented a UV spectra and molecular ion equivalent to 2,3-dihydroxy-1-(4-hydroxy-3-methoxyphenyl)-propan-1-one (Table 1⇑), previously identified in both brandies and wines (Gómez-Cordovés et al. 1997, Peña-Neira et al. 2000, Baderschneider and Winterhalter 2001). Glutathionyl adducts derived from hydroxycinnamic acids, which were previously reported in wines (Ibern-Gómez et al. 2002), were not detected under our conditions.

Stilbenes.

cis- and trans-Resveratrol (peaks 56 and 53) and their respective glucosides (peaks 52 and 39) were identified. In the case of the latter forms, a fragment ion [M-H-162]^- corresponding to resveratrol after the loss of the glucose moiety was observed in addition to the molecular ion (Table 1⇑).

Phenolic alcohols and other related compounds.

Peak 8 corresponded to tyrosol (p-hydroxyphenylethanol), an alcohol formed from tyrosine (3-(4-hydroxyphenyl)-alanine) during yeast fermentation. Peak 4 presented a molecular ion ([M-H]⁻ = m/z 153) 16 amu higher than tyrosol ([M-H]= m/z 137), which is equivalent to an additional hydroxyl group (-OH) and therefore consistent with the structure of a dihydroxyphenylethanol (Table 1⇑). Tryptophol ([M-H]⁻ = m/z 160), a nonphenolic alcohol synthesized from tryptophan (2-amino-3-(3-indolyl)-propionic acid), corresponded to peak 46.

Flavonols.

The aglycones myricetin (51) and quercetin (55) were identified in the different wines. Kaempferol ([M-H]⁻ = m/z 285) and isorhamnetin ([M-H]⁻ = m/z 315) were detected at very high retention times (<100 min) and were not evaluated quantitatively. The derivatives myricetin-3-O-glucuronide (36), myricetin-3-O-glucoside (37), quercetin-3-O-galactoside (41), quercetin-3-O-glucuronide (42), quercetin-3-O-glucoside (44), and kaempferol-3-O-glucoside (49) were also identified. In addition to the molecular ion, the fragment ions [M-H-162]⁻ (for glucose/galactose derivatives) and [M-H-176]⁻ (for glucuronide derivatives) were detected (Table 1⇑). The nature of the hexose was determined by comparison with the retention time of glycosylated commercial standards from the same aglycone under identical chromatographic conditions.

Dihydroflavonols.

Peak 43 corresponded to astilbin (dihydroquercetin-3-O-rhamnoside), a compound recently identified both in white and red wines (Baderschneider and Winterhalter 2001, Landrault et al. 2002).

Flavanols.

Nongalloylated and galloylated monomeric ([M-H]⁻ = m/z 289 and 441, respectively) and dimeric ([M-H] ⁻ = m/z 577 and 729, respectively) flavan-3-ols and trimeric procyanidins ([M-H] ⁻ = m/z 865) were identified in the different wine extracts. (+)-Catechin (15), (−)-epicatechin (28), and epicatechin-3-O-gallate (38) were detected as monomeric flavan-3-ols. The procyanidins identified included the dimers B3 (11), B1 (12), B4 (21), B2 (24), and B5 (35); the trimers T2 (16), T3 (19), and C1 (31); one dimer gallate (peak 34); and two trimers (peaks 29 and 33) of unknown structure (Table 1⇑). Identification was also performed in comparison with the retention time of commercial standards and of grape seed procyanidins previously characterized (Monagas et al. 2003). The fragmentation pathway for oligomeric flavanols under ESI-MS in negative mode consisted of ions derived from: (1) the interflavanic bond cleavage ([M_(base)-H]⁻ , [M_{(middle-base)}-H] ⁻, [M_(top)-3H] ⁻ , [M_(top-middle)-3H] ⁻); (2) the Retro-Diels-Alder (RDA) fission on the C ring followed by the loss of a water molecule ([M-H-152] ⁻ , [M-H-152-H₂O] ⁻); and (3) the elimination of a phloroglucinol molecule ([M-H-126] ⁻), as reported by other authors (Friedrich et al. 2000, Monagas et al. 2003). Only the most abundant ions for each type of procyanidin are reported in Table 1⇑.

In the case of epicatechin-3-O-gallate (38), fragment ions resulting from the cleavage of the ester bond were detected at m/z 289 for the (−)-epicatechin unit and at m/z 169 for the gallic acid moiety (Table 1⇑). The unknown dimer gallate (34) presented the ions corresponding to the loss of a gallic acid moiety (m/z 577) (Table 1⇑), although the ions resulting from the cleavage of the interflavanic bond previous to the loss of the gallic acid moiety (m/z 439 [for epicatechin-3-O-gallate-(epi)catechin] or m/z 441 [for (epi)catechin-epicatechin-3-O-gallate]) were not detected, complicating the final assignment of the galloylated unit within the dimer. Based on its molecular ion and fragment ion resulting from the RDA fission, peak 10 (m/z 593, 425) corresponded to the dimer (epi)gallocatechin-(epi)catechin (Table 1⇑) (De Pascual-Teresa et al. 2000, Friedrich et al. 2000).

Contents and relative distributions of nonanthocyanin phenolic compounds.

In order to apply the information provided by ESI-MS, nonanthocyanin phenolic compounds were quantified in Tempranillo, Graciano, Cabernet Sauvignon, and Merlot wines. Because of the complexity of the chromatograms (Figure 1⇑), the different compounds quantified were selected on the basis of their concentration in wine and their chromatographic resolution. The concentration of nonanthocyanin phenolic compounds, either individual, grouped, or as a total, is presented in Table 2⇓. Since the mean recovery value of phenolic compounds in diethyl ether and ethyl acetate is relatively high (87% as a mean; Fernández de Simón et al. 1990), the data presented refers to the wine extracts.

Of the nonflavonoid phenolics, hydroxybenzoic acids and their derivatives and phenolic alcohols and related compounds were the most abundant subgroups in the different wines, accounting for 16.3 to 29.5% and 9.4 to 17.0% of total nonanthocyanin phenolics quantified, respectively (Table 2⇑). Mayén et al. (1995) also found that hydroxybenzoic acids represented the largest proportion of nonflavonoid phenolic compounds in wines from Tempranillo and Cabernet Sauvignon grape varieties cultivated in the south of Spain. The total concentration of hydroxybenzoic acids and their derivatives was similar among varieties, although Graciano was slightly higher. However, greater differences were found in the subgroup of phenolic alcohols and other related compounds; Cabernet Sauvignon had the highest concentration, especially in tyrosol (Table 2⇑). Gallic acid was the most abundant hydroxybenzoic acid quantified in the wines, as reported by other authors (Ritchey and Waterhouse 1999, Peña-Neira et al. 2000), followed by syringic acid. For gallic acid derivatives, ethyl gallate had a higher concentration than methyl gallate. Quantitative data on these latter compounds in wines is reported here for the first time.

For hydroxycinnamic acids and their derivatives, the percentage distribution was very similar among varieties, although Cabernet Sauvignon wine had the highest total concentration because of its high trans-caftaric acid content, followed by Graciano, Merlot, and Tempranillo (Table 2⇑). The concentration of tartaric esters of hydroxycinnamic acids was generally higher than that of free acids (Ritchey and Waterhouse 1999, Burns et al. 2000). The concentration of hexose esters of trans-p-coumaric acid was similar in all four grape varieties. Concentration of stilbenes was extremely low in Cabernet Sauvignon compared with the other varieties, especially Graciano, which had the highest concentration (Table 2⇑). As expected, the trans- forms of resveratrol-3-O-glucoside and its aglucone were the most abundant ones (Mattivi et al. 1995, Burns et al. 2000).

Flavanols were the principal flavonoid compounds found. They comprised 42.9 to 56.1% of total nonanthocyanin phenolics in the different varieties (Table 2⇑). Graciano and Cabernet Sauvignon had the highest total flavanol concentrations followed by Merlot and then Tempranillo. Other authors (Mayén et al. 1995, Monagas et al. 2003) also found low concentrations of flavanols in Tempranillo wines when compared to other varieties, probably because of the low total flavanol content of Tempranillo seeds. Flavan-3-ol monomers were present in higher concentration than the procyanidin dimers in three wines, with (+)-catechin more abundant than (−)-epicatechin. The exception was Graciano, which exhibited similar levels of both monomers (Burns et al. 2000, Monagas et al. 2003). Graciano and Merlot had similar total concentrations of flavonols, which were higher than in Cabernet Sauvignon and Tempranillo, especially with respect to quercetin-3-O-glucuronide (Table 2⇑).

Graciano and Cabernet Sauvignon wines had the highest total concentration of nonflavonoid, flavonoid, and total nonanthocyanin phenolic compounds quantified, followed by Merlot and Tempranillo (Table 2⇑). Flavonoid compounds represented the highest proportion of total nonanthocyanin phenolics in the four varieties, as reported by other authors (Mayén et al. 1995). In general, the concentration of individual compounds is in the range presented for Spanish wines from different geographical origins (Peña-Neira et al. 2000, Mayén et al. 1995), although lower for some compounds when compared with wines from other countries (Ritchey and Waterhouse 1999, Burns et al. 2000). Since the different grape varieties were cultivated in the same vineyard and the respective wines were vinified under identical conditions, the differences in phenolic profiles could be due to grape variety. However, more wines from the same varieties and from other production areas and vintages should be studied in order to confirm the results found in this work.