Measurement of the Effect of Accelerated Aging on the Aromatic Compounds of Gewürztraminer and Teroldego Wines, Using a SPE-GC-MS/MS Protocol

Knowing in detail how the white and red wine aroma compounds behave under various storage conditions and especially at high temperature is important in order to understand the changes occurring to their sensorial character during the shelf life. The initial aim of this work was to develop and validate a fast, modern, robust, and comprehensive protocol for the quantification of 64 primary, secondary, and tertiary volatile compounds by using solid-phase extraction (SPE) cartridges in sample preparation and fast GC-MS/MS (gas chromatography-tandem mass spectrometry assay) in analysis. The protocol was applied to a study of the behavior of seven Gewürztraminer and seven Teroldego wines stored in anoxia at 50 °C for 2.5 and 5 weeks. The results demonstrated a sharp decrease of the main linear terpenes linalool, geraniol, and nerol and the consequent increase of the cyclic ones, such as α-terpineol and 1,8-cineole; the increase of the C13-norisoprenoids 1,1,6,-trimethyl-1,2-dihydronapthalene (TDN), and β-damascenone and the C10 norisoprenoid safranal; the hydrolysis of acetates and linear esters; and the increase of some branched-chain esters. In red wines, a moderate increase was observed for some lactones. Some unwanted compounds, such as 2-aminoacetophenone (2-AAP), showed a notable increase in some Gewürztraminer wines, exceeding the olfactory threshold.


Introduction
The analysis of volatile compounds in wine is an informative tool for characterizing the different cultivars and wine styles and for studying their sensory properties and the dynamic evolution of their composition during maturation and aging. Indeed, we know that wine is one of the beverages that can often evolve and improve during the maturation phase between the fermentation and the bottling as well as during the aging in bottle if this is done in optimal conditions [1,2].
The complete analysis of the wine aroma is, however, complex, time consuming, and expensive. The concentration of the key compounds contributing to the aroma of wines has an extremely wide range of concentration (ng-mg/L) and equally diverse chemical characteristics that sometimes require specific and selective detection methods [3]. The main classes of compounds that impact the fruity and flowery aroma of wines and that modify it The results obtained from the various extractions showed that in all the 3 cartridges, part of the first 1.3 mL DCM fraction remained trapped into the resin. However, almost half remained in the Bond Elut ENV cartridge, and some water was also retained. These cartridges were also found to be less efficient for the extraction of alcohols and some esters, and for all these reasons, the Bond Elut ENV was excluded ( Figure S1, Table S1).
The other two cartridges, Isolute ® ENV+ and LiChrolut ® EN, had very similar performance. However, while the experiment was ongoing, we learned that the latter will soon be removed from the market, so we decided to further validate the method with the Isolute ® ENV+ cartridges. Considering that in these cartridges, too, a small amount of compounds was found in the second DCM fraction, it was decided to elute with 2 mL instead of 1.3 mL of DCM. To evaluate the repeatability of the method, technical replicates were made within one day (intraday) and between-day (interday) using both white and red wine mixes. Repeatability (Supplementary Table S2) of the extraction resulted in a CV% below 10% for most compounds (n = 70). For two compounds, the CV% gave values between 10% and 20%, which were still acceptable. Only 2 compounds, acetoin (intraday and interday) and phenylacetaldehyde (interday), in the red wine samples, gave values over 20% and therefore were excluded from the method. For white wine, all the CV% values were below 16%. R2 was in a range from 0.9907 to 0.9999 for all compounds and indicated good fit and linearity for the calibration curves in relation to the scope of the method. The results obtained from the various extractions showed that in all the 3 cartridges, part of the first 1.3 mL DCM fraction remained trapped into the resin. However, almost half remained in the Bond Elut ENV cartridge, and some water was also retained. These cartridges were also found to be less efficient for the extraction of alcohols and some esters, and for all these reasons, the Bond Elut ENV was excluded ( Figure S1, Table S1).
The other two cartridges, Isolute ® ENV+ and LiChrolut ® EN, had very similar performance. However, while the experiment was ongoing, we learned that the latter will soon be removed from the market, so we decided to further validate the method with the Isolute ® ENV+ cartridges. Considering that in these cartridges, too, a small amount of compounds was found in the second DCM fraction, it was decided to elute with 2 mL instead of 1.3 mL of DCM. To evaluate the repeatability of the method, technical replicates were made within one day (intraday) and between-day (interday) using both white and red wine mixes. Repeatability (Supplementary Table S2) of the extraction resulted in a CV% below 10% for most compounds (n = 70). For two compounds, the CV% gave values between 10% and 20%, which were still acceptable. Only 2 compounds, acetoin (intraday and interday) and phenylacetaldehyde (interday), in the red wine samples, gave values over 20% and therefore were excluded from the method. For white wine, all the CV% values were below 16%. R2 was in a range from 0.9907 to 0.9999 for all compounds and indicated good fit and linearity for the calibration curves in relation to the scope of the method.
Most of the compounds (n = 48) gave optimal recovery values between 80-120%, and 13 compounds gave a recovery between 60-80%. Only a dozen compounds in both red and white wines gave values <50%; these were mostly high polar compounds, which are unable to bind to the non-polar stationary phase of styrene divinylbenzene, or acid compounds, for which the pH of the matrix should be changed, with the risk of losing other compounds of interest. For some compounds present in large quantities, such as ethyl esters, diethyl succinate, octanoic acid, decanoic acid, and benzyl alcohol, we tried to increase the split ratio in the GC injector from 1:10 to 1:150, obtaining better results. Both in red and white wines, the recovery values of menthalactone thus improved, probably due to a reduction of the baseline in the chromatogram. However, considering that the 1:10 split ratio is better for the vast majority of compounds, it was decided to use that injection condition and to inject with the highest splitting ratio (1:150) only to quantify the compounds present at higher concentration (Table S2). The limits of quantification (LOQ) for all compounds were suitable for their quantification both in red and white wines. The linearity for the major compounds could be increased using the highest splitting ratio (1:150). The chromatographic run of only 16 min allows a high production capacity. The extraction method, together with the fast GC-MS/MS analysis, made it possible to significantly reduce the use of the DCM solvent, with advantages in terms of operator safety as well as time, avoiding further concentrations of the extracts and allowing the quantification of 64 compounds. All the validation parameters are reported in Tables 1 and S2. This validated method was used to monitor the behavior of volatile compounds in Gewürztraminer wines and in autochthonous red wines of the Teroldego variety during an accelerated aging period, and the results are reported in Tables 2 and 3.     Compounds  T1  T2  T3  T4  T5  T6  T7  T1  T2  T3  T4  T5  T6  T7  T1  T2  T3  T4  T5  T6  T7 isobutyl   Compounds  T1  T2  T3  T4  T5  T6  T7  T1  T2  T3  T4  T5  T6  T7  T1  T2  T3  T4  T5  T6

Accelerated Aging
A small experiment was carried out to evaluate the repeatability of the accelerated aging method. Five technical replicates of two different Gewürztraminer wines were placed for 4 days at 40 • C and then analyzed. The results are shown in Table S3 and show, for all compounds, the CV% is below 16%. In consideration of these results, it was decided to conduct the experiment using seven biological replicas of commercial wines for both Gewürztraminer and Teroldego. The wines were analyzed at time 0 (t0) and after 2.5 (t1) and 5 weeks (t2) at 50 • C following the method proposed by Ferreira [14]. The oxygen content was also measured during the first week of storage. The dissolved oxygen content at time zero in the different samples was very variable, also depending on the type of cap used, but always under 578 ppb for red wines and under 604 ppb in white wines. It was also seen that already after 2-3 days, the concentration was very low, under 50 ppb for all the samples.

Gewürztraminer Wine
This aromatic variety with scent of rose petals, cloves, lychees, and other tropical fruits is a variety widely cultivated in the Trentino Alto Adige region located in northern Italy, especially in the area of Tramin, and it has long been studied by various researchers to try to understand what the most characterizing components are [15][16][17].
Terpenes and monoterpenols, particularly geraniol, cis rose oxide, citronellol, and linalool, are responsible for the characteristic floral aroma in the Vitis vinifera cv. Gewürztraminer grapes and wines [16]. During wine processing and aging, many acid-catalyzed rearrangements take place, mainly with an increase in cyclic forms or hydroxylated derivatives, and this involves changes in concentration and the formation of new volatile compounds that were not present in the grapes or in young wines. Usually the open-chain monoterpene alcohols have a lower perception threshold than their cyclic equivalents, and this accounts for the reduction in the typical floral aroma during storage or aging (reference). The data analysis of the measurements demonstrated a substantial decrease of the mean values by 79% for linalool, by 92% for nerol, by 93% for geraniol, and by 78% for citronellol ( Figure 2). The one-way ANOVA analysis pointed out that this change was statistically significant already after 2.5 weeks of accelerated aging (Table S4). On the contrary, the mean values of cyclic α-terpineol and terpinen 4-ol showed a statistically significant increment the first 2.5 weeks (from t0 to t1), correspondingly from 90.14 µg/L to 319.94 µg/L and from 2.47 µg/L to 8.28 µg/L ( Figure 2 and Table S4). It is interesting to note that the 1,8-cineole content increased considerably, going from 0.04 µg/L to 3.57 µg/L, and in the two wines with linalool content above 300 µg/L, which also had a higher concentration of α-terpineol, this compound was present in an amount of more than 5 µg/L at the end of the 5 weeks; this supports the theory that this compound may form from linalool cyclization reactions (Figure 2) [4]. Furthermore, the increase of 1,8-cineole was statistically significant after 2.5 weeks if we consider the mean value of the seven biological replicates (Table S4). Previous studies showed that α-terpineol can be formed from limonene under acidic conditions but could also derive from the cyclization of geraniol, nerol, and linalool; after that, α-terpineol can be transformed directly into 1,8-cineole or into 1,8-terpine and this latter compound to 1,8-cineole [4]. 1,8-cineole, with a eucalyptus odor and a very low threshold of around 2 µg/L [18,19], could contribute to the wine aging aroma. Pyranic oxides of linalool were among the compounds that increased during the study, probably due to hydrolysis from the bound forms (+1322%). During fermentation and aging, the aglycones should be freed from precursors; however, in this experiment, probably due to the too-high temperature, no initial increase was observed, while there was a decrease. It is assumed that they undergo rearrangement towards more stable cyclic forms very quickly after the hydrolysis.  (Table S4)).
Norisoprenoids are among the most important evolutionary wine aroma pounds; they can be formed by direct degradation of carotenoids, such as β-caroten neoxanthin, or they can be stored as glycoconjugates, which can then release their v aglycone during fermentation or aging via enzymatic and acid hydrolysis processe carotenoid content in grapes, the fermentation process and the wine storage cond are factors that greatly influence the evolutionary profile of the wine [20]. One of th important norisoprenoids is certainly β-damascenone that, with its very low thresho ng/L), manages to contribute to the aroma of the wine both directly and indirectly enhancer of the fruity note [21]. During the experiment (Figure 3), the mean content compound increased slightly but was statistically significant after 2.5 weeks, from µg/L to 3.32 µg/L. The TDN mean content increased statistically significant alread 2.5 weeks, too, but was considerably more by up to 12 times, from 0.87 to 13.19 µg/L ure 3 and Table S4). Such a behavior is in accordance with the literature since the pr tion of TDN in wine is promoted by the heating [22][23][24].
Another compound that increased greatly with heating was safranal, (2,6,6thyl-1,3-cyclohexadiene-1-carboxaldehyde), which went from 0.14 µg/L at t0 to 1.1 at t2. This increase was statistically significant both between t0 and t1, and t1 and t2 S4). Safranal is the main aroma component of saffron; in wine, it exists in free form but given its considerable increase after heating and also in light of the observed in of its concentration in reserve sparkling wines [25], it is possible to confirm the pr of some precursors. In saffron, the main monoterpene glycoside precursor of safr picrocrocin [26]. β-glucosidase action, thermal treatment, or alkaline-acid hydroly picrocrocin liberate the aglycone directly or enzymatically, with the formation of droxy-2,6,6-trimethyl-1-cyclohexene-1-carboxaldehyde (HTCC, C10H16O2), which is formed to safranal by dehydration during the drying process of the plant material was also reported that crocetin dialdehyde could be oxidized and esterified to ge crocetin esters, which could also be a safranal precursor after an enzymatic or th treatment [28]. In our samples, however, in the untreated wine samples (t0), we d find any picrocrocin peak, which can therefore lead to the hypothesis that safra formed starting from some other precursor or by the rearrangement of some other cules.  (Table S4)).
Norisoprenoids are among the most important evolutionary wine aroma compounds; they can be formed by direct degradation of carotenoids, such as β-carotene and neoxanthin, or they can be stored as glycoconjugates, which can then release their volatile aglycone during fermentation or aging via enzymatic and acid hydrolysis processes. The carotenoid content in grapes, the fermentation process and the wine storage conditions are factors that greatly influence the evolutionary profile of the wine [20]. One of the most important norisoprenoids is certainly β-damascenone that, with its very low threshold (50 ng/L), manages to contribute to the aroma of the wine both directly and indirectly as an enhancer of the fruity note [21]. During the experiment (Figure 3), the mean content in this compound increased slightly but was statistically significant after 2.5 weeks, from 2.39 µg/L to 3.32 µg/L. The TDN mean content increased statistically significant already after 2.5 weeks, too, but was considerably more by up to 12 times, from 0.87 to 13.19 µg/L ( Figure 3 and Table S4). Such a behavior is in accordance with the literature since the production of TDN in wine is promoted by the heating [22][23][24]. . Figure 3. Behavior of the main norisoprenoids in the Gewürztraminer wines stored in anoxia at time zero (t0) and for 2.5 (t1) and 5 (t2) weeks at 50 °C. (Tukey's HSD: t0 a, t1 b, and t2 b for TDN; t0 a, t1 ab, and t2 b for β-damascenone; and t0 a, t1 b, and t2 c for safranal (Table S4).
Esters and Acetates: the behavior of acetates and linear ethyl esters, as widely demonstrated in literature, includes a rapid decrease during aging, especially if the wine is not stored in suitable conditions [29]. Figure 4 shows that hexyl acetate decreased rather quickly, going from an average value of 71.55 µg/L to 26.30 µg/L after 2.5 weeks and to 14.09 µg/L at the end of the experiment. Isobutyl acetate showed a similar trend, from the initial concentration of 46.15 µg/L to the final one of 24.85 µg/L (Table S4). Octanoic and decanoic ethyl esters also decreased by 63% and 85%, respectively. In the family of fruity esters, the ethyl esters of the branched acids followed a completely different aging pattern compared to linear ethyl esters and acetates. The levels of these esters progressively increased during aging in a statistically significant way (Table S4). Ethyl 2-methylbutyrate and ethyl 3-methylbutanoate (ethyl isovalerate) exhibited the opposite behavior and increased by +168% and 182%, respectively. Ethyl 2-hydroxy-4-methylpentanoate (ethyl leu-  HSD: t0 a, t1 b, and t2 b for TDN; t0 a, t1 ab, and t2 b for β-damascenone; and t0 a, t1 b, and t2 c for safranal (Table S4).
Another compound that increased greatly with heating was safranal, (2,6,6-trimethyl-1,3-cyclohexadiene-1-carboxaldehyde), which went from 0.14 µg/L at t0 to 1.13 µg/L at t2. This increase was statistically significant both between t0 and t1, and t1 and t2 (Table  S4). Safranal is the main aroma component of saffron; in wine, it exists in free form [25], but given its considerable increase after heating and also in light of the observed increase of its concentration in reserve sparkling wines [25], it is possible to confirm the presence of some precursors. In saffron, the main monoterpene glycoside precursor of safranal is picrocrocin [26]. β-glucosidase action, thermal treatment, or alkaline-acid hydrolysis on picrocrocin liberate the aglycone directly or enzymatically, with the formation of 4-hydroxy-2,6,6-trimethyl-1-cyclohexene-1-carboxaldehyde (HTCC, C 10 H 16 O 2 ), which is transformed to safranal by dehydration during the drying process of the plant material [27]. It was also reported that crocetin dialdehyde could be oxidized and esterified to generate crocetin esters, which could also be a safranal precursor after an enzymatic or thermal treatment [28]. In our samples, however, in the untreated wine samples (t0), we did not find any picrocrocin peak, which can therefore lead to the hypothesis that safranal is formed starting from some other precursor or by the rearrangement of some other molecules.
Esters and Acetates: the behavior of acetates and linear ethyl esters, as widely demonstrated in literature, includes a rapid decrease during aging, especially if the wine is not stored in suitable conditions [29]. Figure 4 shows that hexyl acetate decreased rather quickly, going from an average value of 71.55 µg/L to 26.30 µg/L after 2.5 weeks and to 14.09 µg/L at the end of the experiment. Isobutyl acetate showed a similar trend, from the initial concentration of 46.15 µg/L to the final one of 24.85 µg/L (Table S4). Octanoic and decanoic ethyl esters also decreased by 63% and 85%, respectively. In the family of fruity esters, the ethyl esters of the branched acids followed a completely different aging pattern compared to linear ethyl esters and acetates. The levels of these esters progressively increased during aging in a statistically significant way (Table S4). Ethyl 2-methylbutyrate and ethyl 3-methylbutanoate (ethyl isovalerate) exhibited the opposite behavior and increased by +168% and 182%, respectively. Ethyl 2-hydroxy-4-methylpentanoate (ethyl leucate) was identified for the first time in red and white table wines as a compound directly associated with a "fresh blackberry" aroma [30]. This ester increased by +96% between the beginning and the end of the experiment. Aging would seem to favor an increase in the overall concentration of ethyl leucate [9] since the acid-ester equilibrium was the most effective in generating the branched fatty acid ethyl esters from their corresponding acids during wine aging [31]. Diethyl succinate (+60%), as already reported for the esters of diprotic acids, increased during aging and were sometimes used as aging markers [32,33].
Esters and Acetates: the behavior of acetates and linear ethyl esters, as widely demonstrated in literature, includes a rapid decrease during aging, especially if the wine is not stored in suitable conditions [29]. Figure 4 shows that hexyl acetate decreased rather quickly, going from an average value of 71.55 µg/L to 26.30 µg/L after 2.5 weeks and to 14.09 µg/L at the end of the experiment. Isobutyl acetate showed a similar trend, from the initial concentration of 46.15 µg/L to the final one of 24.85 µg/L (Table S4). Octanoic and decanoic ethyl esters also decreased by 63% and 85%, respectively. In the family of fruity esters, the ethyl esters of the branched acids followed a completely different aging pattern compared to linear ethyl esters and acetates. The levels of these esters progressively increased during aging in a statistically significant way (Table S4). Ethyl 2-methylbutyrate and ethyl 3-methylbutanoate (ethyl isovalerate) exhibited the opposite behavior and increased by +168% and 182%, respectively. Ethyl 2-hydroxy-4-methylpentanoate (ethyl leucate) was identified for the first time in red and white table wines as a compound directly associated with a "fresh blackberry" aroma [30]. This ester increased by +96% between the beginning and the end of the experiment. Aging would seem to favor an increase in the overall concentration of ethyl leucate [9] since the acid-ester equilibrium was the most effective in generating the branched fatty acid ethyl esters from their corresponding acids during wine aging [31]. Diethyl succinate (+60%), as already reported for the esters of diprotic acids, increased during aging and were sometimes used as aging markers [32,33]. Phenols: a very important compound for the spicy note of Gewürztraminer is 4-vinylguaiacol, which brings clove notes and is often present in quantities much higher than its olfactory threshold (40 µg/L) [34]. The behavior of this compound during aging is well Phenols: a very important compound for the spicy note of Gewürztraminer is 4-vinylguaiacol, which brings clove notes and is often present in quantities much higher than its olfactory threshold (40 µg/L) [34]. The behavior of this compound during aging is well known: it tends to decrease rapidly, with the half-life of vinylphenol in white wines being approximately 6 months at 16-18 • C [35]. It was found that the main degradation product of 4-vinylguaiacol in beer was apocynol (4-(1-hydroxyethyl)-2-methoxyphenol) [36], while another possibility is that 4-vinylguaiacol could react with ethanol to form ethoxyethyl phenols, as observed in some wines [37]. In our case, we observed a statistically significant loss of 56% of 4-vinylguaiacol in 5 weeks at 50 • C (Table S4).
Other important benzenoids: methyl salicylate is an organic ester naturally produced by many plants, particularly wintergreens, and also present in wine, sometimes in quite high quantities, such as in the Verdicchio and Lugana varieties [38,39]. It was demonstrated that it could be present in both free and bound form (MeSA glycosides). In small quantities, it is also present in Gewürztraminer, and during the experiment, the content increased a little due to hydrolysis by the glycosides although it remained very far from the olfactory threshold (50 µg/L) and is not statistically significant. 2-Aminoacetophenone (2-AAP) is a known compound since it is considered the main cause of the so-called untypical "aging off-flavor" (UTA) in Vitis vinifera wines. According to the literature, the formation of 2-AAP is caused by the oxidative degradation of the phytohormone indole-3-acetic acid (IAA) after fermentation. 2-AAP was identified as the character impact compound, with an odor threshold of about 1 µg/L in wine by [40,41]. In this experiment (Figure 4), the initial value of 0.23-0.27 µg/L of this compound was very similar for all the wines, while at the end of the 5 weeks, in three wines, it had increased to very close (0.81-0.93 µg/L) to the sensory threshold.

Teroldego Wines
Teroldego is a red autochthonous variety from the Trentino-Alto Adige region in northern Italy, and despite their dark color, Teroldego grapes produce wines that have bright fruity notes.
Other important benzenoids: methyl salicylate is an organic ester naturally pro by many plants, particularly wintergreens, and also present in wine, sometimes in high quantities, such as in the Verdicchio and Lugana varieties [38,39]. It was d strated that it could be present in both free and bound form (MeSA glycosides). In quantities, it is also present in Gewürztraminer, and during the experiment, the c increased a little due to hydrolysis by the glycosides although it remained very fa the olfactory threshold (50 µg/L) and is not statistically significant. 2-Aminoacetoph (2-AAP) is a known compound since it is considered the main cause of the so-call typical "aging off-flavor" (UTA) in Vitis vinifera wines. According to the literatu formation of 2-AAP is caused by the oxidative degradation of the phytohormone i 3-acetic acid (IAA) after fermentation. 2-AAP was identified as the character impac pound, with an odor threshold of about 1 µg/L in wine by [40,41]. In this experimen ure 4), the initial value of 0.23-0.27 µg/L of this compound was very similar for wines, while at the end of the 5 weeks, in three wines, it had increased to very close 0.93 µg/L) to the sensory threshold.

Teroldego Wines
Teroldego is a red autochthonous variety from the Trentino-Alto Adige reg northern Italy, and despite their dark color, Teroldego grapes produce wines tha bright fruity notes.
Norisoprenoids: In this red wine as well, TDN, β-damascenone, and safranal increased a great deal during the experiment due to hydrolysis/rearrangement from its precursors ( Figure 7).
2-AAP: in Teroldego wines, the content of 2-AAP did not increase during the experiment; in fact, the aging off-flavor (UTA) has not yet been detected in red wines, and red wines spiked with the precursor indole-3-acetic acid before fermentation did not show any significant formation of 2-AAP [42,43]. In fact, red wines are far richer in polyphenols than white wines, which are able to protect wine from oxidation, including the reactions driving to the release of 2-AAP.  (Table S5)).

Chemicals and Reagents
All standards used in this study are listed in Table S2. Ethanol 99.8%, n-heptanol 99.9%, dichloromethane 99.8%, and methanol for HPLC 99.9% were purchased from Sigma-Aldrich (St. Luis, MO, USA); 3 cartridges with 200 mg of stationary phase based on  (Table S5)).
Norisoprenoids: In this red wine as well, TDN, β-damascenone, and safranal increased a great deal during the experiment due to hydrolysis/rearrangement from its precursors (Figure 7).  (Table S5)).
Norisoprenoids: In this red wine as well, TDN, β-damascenone, and safranal increased a great deal during the experiment due to hydrolysis/rearrangement from its precursors ( Figure 7).
2-AAP: in Teroldego wines, the content of 2-AAP did not increase during the experiment; in fact, the aging off-flavor (UTA) has not yet been detected in red wines, and red wines spiked with the precursor indole-3-acetic acid before fermentation did not show any significant formation of 2-AAP [42,43]. In fact, red wines are far richer in polyphenols than white wines, which are able to protect wine from oxidation, including the reactions driving to the release of 2-AAP. Figure 7. Behavior of the main norisoprenoids in the Teroldego wines stored in anoxia at time zero (t0) and for 2.5 (t1) and 5 (t2) weeks at 50 °C. (Tukey's HSD: TDN and β-damascenone: t0 a, t1 b, and t2b and for safranal: t0 a, t1 b, and t2 c (Table S5)).
2-AAP: in Teroldego wines, the content of 2-AAP did not increase during the experiment; in fact, the aging off-flavor (UTA) has not yet been detected in red wines, and red wines spiked with the precursor indole-3-acetic acid before fermentation did not show any significant formation of 2-AAP [42,43]. In fact, red wines are far richer in polyphenols than white wines, which are able to protect wine from oxidation, including the reactions driving to the release of 2-AAP.

Wine Samples
Ten different wines varieties (five white and five red) were blended to create a representative white and red matrix to be used for the optimization of the extraction method. For the accelerated aging, 7 different commercial Gewürztraminer wines of the 2019 vintage and 7 different commercial Teroldego wines of the 2019 vintage were acquired from different wineries in Trentino Alto Adige region. The basic enological analysis can be found as Supplementary in Table S6.

Wine Aging
The wine bottles were opened under a N 2 hood and aliquoted in two technical replicates into 50-mL clear glass bottles, avoiding any headspace, and then, the bottles were enclosed in vacuum bags. Internally to each bottle was placed a Pst3 oxygen sensor (Nomacorc SA, Thimister-Clemont, Belgium) to measure the dissolved oxygen, which was also the total packaging oxygen (TPO), because the bottles were full. For the accelerated aging, the samples were stored at 50 • C in a laboratory heater. Each wine sample was analyzed immediately after 2.5 (first replicate) and 5 (second replicate) weeks of accelerated aging. Since the oxygen sensors were placed internally, the measurement was carried out using luminescence technology optical fibre outside the glass bottle by using the NomaSense system (Nomacorc SA, Thimister Clemont, Belgium).

Sample Preparation and Extraction
Sample preparation and extraction of the free aroma compounds was performed according to the modification of the method described in [9,44]. Solid-phase extraction was initially performed using 3 different cartridges, namely Bond Elut ENV (Agilent, Santa Clara, CA, USA), Isolute ® ENV+ (Biotage, Uppsala, Sweden), and LiChrolut ® EN (Merk, Darmstadt, Germany), filled with 200 mg stationary phase and pre-conditioned with 4 mL dichloromethane, followed by 4 mL of methanol and 4 mL of model wine. A total of 50 mL of wine mixed with 100 µL of internal standard (n-heptanol 250 mg/L) was loaded onto the cartridge, which was then washed with 3 mL of water. The cartridges were dried for 10 min and tested as reported above. The validated method uses the Isolute ® ENV+ cartridge that was pre-conditioned, loaded and dried in the same way, and eluted with 2 mL dichloromethane directly into the injection vials.

MS-MS Optimization
The list of compounds was put together in order to include the most important chemical classes (varietals, fermentative, and aging) for wine aroma. The optimization of the MS/MS method was performed for all compounds, diluted in ethanol solution, and injected in EI and operated in MRM mode. The optimizer software (embedded in MassHunter Workstation) was used in order to acquire two or three MS/MS transitions and after that to select the best collision energy for each transition. The results with all the settings parameters are reported in Table 1.

GC-MS/MS Analysis
The instrument method was adapted from [10] with some modification, using the Agilent Intuvo 9000 system for fast GC coupled with an Agilent 7010B triple quadrupole mass spectrometer (Agilent Technologies, Santa Clara, CA, USA) equipped with an electronic ionization source operating at 70 eV. The separation was achieved by injecting 1 µL in split mode (1:10) into a DB-Wax Ultra Inert column (30 m × 0.25-mm id × 0.25-µm film thickness, Agilent Technology, Santa Clara, CA, USA). The initial temperature of the GC oven was 40 • C for 2 min, ramped up by 10 • C/min to reach 55 • C, then by 20 • /min until 165 • C, by 40 • C/min to 240 • C for 1.5 min, and finally by 50 • C/min to 250 • C and kept at this temperature for additional 4 min (16 total runtime). Helium was used as carrier gas (with a flow of 1.2 mL/min). The mass spectra were acquired in multiple reaction monitoring mode. Nitrogen was used as the collision gas, with a flow of 1.5 mL/min in addition with Helium at 4.0 mL/min as quench gas. The transfer line and source temperature were set at 250 • C and 230 • C, respectively. The data acquisition and subsequent analyses were done using the MassHunter Workstation software.

Method Validation
Validation of the extraction and GC/MS/MS method was performed in terms of limit of detection, limit of quantification, linearity range, and inter-and intraday precision (Supplementary Table S2).
The limit of quantification was taken as the lowest point of the calibration curve, and the limit of detection was set at 1/3 times the LOQ. Linearity was studied by injecting each compound at different ranges for a total of 20 concentration points. A calibration curve was established for each of the 64 compounds. The linear calibration parameters were obtained using the least squares regression method. The squared correlation coefficient (R2) was used to estimate linearity. The precision of the method was determined by calculating the coefficient of variation (CV) for daily (intraday) and day-to-day (interday) analysis using the medium spike level and the retention time. The recovery was tested using 3 different spike-level (low, medium, and high) standard solutions. Concentrations were referred to 2 mL in vial. The calculation was expressed by the following formula: Recovery% = [((spiked wine) − wine)/(solvent + spike)] × 100.

Statistical Analysis
The descriptive and ANOVA statistical analysis, and the visualization of the results was performed using SPSS V28 (IBM Statistics). The one-way ANOVA analysis was performed to compare the three groups' means (t0, time zero; t1, 2.5 weeks; and t2, 5 weeks) for each measured compound. For the post hoc multiple comparison, the Tukey's HSD statistical analysis were performed considering as a hypothesis with a p-value less than 0.05.

Conclusions
The study made it possible to identify the best cartridge to allow the main volatile compounds present in wine to be extracted repetitively and accurately. It then made it possible to reduce the volumes of solvents necessary for the preparation of the sample considerably and to elute directly into the vial for injection, avoiding any concentration step. The use of a triple quadrupole also made it possible to reduce the analysis time. Using this method, seven white and seven red wines were analyzed before and after accelerated aging. The analysis allowed us to monitor the behavior of the most important classes of volatile compounds that change during aging, finding many confirmations, such as the hydrolysis of non-volatile glycosidic precursors as well as the chemical rearrangements of certain terpene compounds with the formation of new impact molecules that are sometimes very important for aging red wine aroma; it will be necessary to test whether these notes are also appreciated in aromatic white wines. For other compounds, the analysis confirmed the already well-known behavior: the hydrolysis of acetates and linear ethyl esters, with consequent loss of fruity notes and the increase of some branched esters, which, especially in red wines, support the fruity note. New observations that will need to be explored also emerged, such as the high increase in safranal, a C10 norisoprenoid, during aging. The precursor of this compound in wine is not already known. From the results, it is also evident that many compounds reached the maximum quantity already after 2.5 weeks at 50 • C; however, studies at lower temperatures will be necessary to better understand these trends. The preliminary results obtained in the experiments of accelerated aging are promising and suggest that the method here employed could represent an affordable analytical tool in the quest to predict the aromatic potential during aging. We are aware that further work is needed, but a step has been made towards the validation of a protocol that could support winemakers in the selection of the wine lots suitable to produce reserve wine.

Supplementary Materials:
The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/metabo12020180/s1, Figure S1: Elution tests from cartridges MI (1st extraction) and MII (2nd extraction) with 3 aliquots of DCM solvent (1, 2, 3) Cartridge B = Bond Elut ENV; I = Isolute® ENV+; L = LiChrolut® EN; W = white wine; M = medium spike, Table S1: Comparison of cartridges; percentage of compounds found in the 2nd and 3rd solvent fractions (dichloromethane) considering 100% the content of the 1st fraction. Descriptive statistics, one-way Anova analysis and post-hoc test (Tukey test p < 0.05) (n = 7 wine sample; 4 white and 3 red), Table S2: Recovery, intraday and interday precision for red and white wines, in red compounds with unacceptable values, Table S3: Repeatability of the accelerated aging treatment of two different Gewürztraminer wines (A, B) kept for 4 days at T equal to 40 • C and then analyzed following the validated SPE-GC-MS/MS method, Table S4: Descriptive statistics of the measured volatile compounds and one-way Anova analysis and post-hoc test results for the Gewürztraminer wines, Table S5: Descriptive statistics of the measured volatile compounds and one-way Anova analysis results for the Teroldego wines, Table S6: basic enological analysis of commercial wines used for accelerating aging.