Approaches to the classification of wine aroma ageing potential. Applications to the case of terpenoids in Valpolicella red wines

Valpolicella is an Italian wine-producing region famous for premium red wines. Valpolicella wines often undergo a minimum ageing period between 1 and 4 years, according to the wine type. Therefore, identifying relevant types of aroma precursors and the related chemical mechanisms behind aroma development during ageing is of great importance for Valpolicella wines. This study assessed free and glycosidically-bound terpene compounds of Valpolicella wines as potential precursors of different ageing-related terpenes using a model ageing experimental approach. Results showed that terpene profiles changed substantially with ageing. The occurrence of cyclic terpenes ( p -menthane-1,8-diol, 1,8-cineole and 1,4-cineole) was mostly associated with the content of free monoterpenols, linalool in particular, in young wines, whereas glycosylated monoterpenols appeared to play a minor role. Accumulation of cineoles was strongly modulated with pH; in fact, low pH values promoted higher cineole yields.


INTRODUCTION
Valpolicella is a region located in the northeast of Italy, in the province of Verona, where the red wines Valpolicella Superiore, Ripasso, Recioto and the most famous Amarone are produced.The grape varieties used to produce these wines are mainly Corvina and Corvinone and a range of minor varieties (MIPAAF, 2019).Valpolicella wines are characterised by cherry, pepper, spicy, tobacco and balsamic notes, among other aromas.(Accordini, 2013;Bellincontro et al., 2016;D'Agata, 2019).Depending on the wine type, a cellar ageing period comprised of one and four years is required by Protected Designation of Origin (PDO) regulation, after which the wine can be released on the market.During ageing, the sensory characteristics of red wine, including its aroma, undergo dramatic changes, with a decrease in fresh fruity attributes and an increase of more complex ripe fruit, dry fruit, tobacco and balsamic aromas (Rapp and Mandery, 1986;McKay et al., 2010;Moio, 2016;Ugliano, 2013).Concerning the latter attributes, recent studies on Valpolicella wines highlighted the potential relevance of different terpenes and norisoprenoids (Slaghenaufi and Ugliano, 2018).While some chemical mechanisms underlying sensory evolution are well known (Antalick et al., 2014;Díaz-Maroto et al., 2005;Tarasov et al., 2020;Skouroumounis and Sefton, 2000), such as esters hydrolysis, acid-catalysed formation non-megastigmane norisoprenoids or oxidation-driven accumulation of aldehydes, others are still poorly understood.Among these, in the case of wines from non-aromatic varieties such as Valpolicella reds, the relative contribution of different forms of various terpenoids and norisoprenoids (e.g., free or glycosylated) requires further investigation (Slaghenaufi et al., 2019).
Among the terpenes potentially contributing to red wine aroma, various floral monoterpenes, as well as the balsamic cyclic terpenes 1,4-cineole and 1,8-cineole, have been reported (Atalick et al., 2015).In the case of Valpolicella wines, some of these were shown to vary significantly according to grape geographical origin, variety and fermentation practices (Slaghenaufi et al., 2019;Luzzini, 2021a;Luzzini et al., 2021b;Luzzini et al., 2021c).In wines from non-aromatic varieties, monoterpene alcohols are largely formed during fermentation from the enzymatic hydrolysis of glycosidic precursors (Ugliano et al., 2006;Slaghenaufi et al., 2019).Conversely, cyclic terpenes, such as cineoles, appear to be mostly associated with ageing (Fariña et al., 2005;Slaghenaufi and Ugliano, 2018), and studies in model systems have shown the existence of complex chemical relationships between these compounds and monoterpene alcohols at wine pH (Slaghenaufi et al., 2019).
At the same time, acid-catalysed hydrolysis of glycosidic precursors can release various terpene compounds some of which can undergo further degradation reactions at wine pH (Skouroumounis and Sefton, 2000).Overall, the terpene content of aged red wine is given by the balance between acid-driven formation and degradation reactions.The formation of a carbocation intermediate plays a central role in these reactions, giving rise to a wide range of reaction products with different aromatic characteristics (Wedler et al., 2015).Some authors have recently linked the occurrence of balsamic notes in aged wines to the formation of cineoles from terpene precursors (Slaghenaufi and Ugliano, 2018;Poitou et al., 2017;Antalick et al., 2015;Fariña et al., 2015).This work aimed to investigate the evolution of the Valpolicella wines terpenes pool during ageing, focusing on the compositional factors of young wines determining the chemical patterns of aged wines.

Wine samples
Wines of the two main grape varieties of the Valpolicella appellation, Corvina and Corvinone, were used in the study.The wines were obtained from grapes harvested in mid-September during the 2017, 2018 and 2019 vintages in five different vineyard parcels belonging to the same winery and located in the Valpolicella region.The same five parcels were samples during the three vintages.A total of 30 wines, 15 Corvina (W1-W15) and 15 Corvinone (W16-W30), were obtained for this study.All vineyards had the same clone-rootstock combination.Grapes were harvested and vinified from 13 to 24 September in 2017, 17 September to 1 October in 2018 and 25 September to 14 October 2019.Harvest took place following the schedule established by the winery owning the vineyards so that grapes were processed in agreement with standard winery evaluation of maturity.After manual destemming, berries were randomised to obtain batches of 7 kg each.From each batch, 3.5 kg were taken, hand crushed with 100 mg/kg of potassium metabisulphite and put into a 5 L glass vessel.Fermentations of fresh grapes musts were carried out in duplicate with Saccharomyces cerevisiae AWRI 796 (AB Mauri, Camellia, Australia).Active dry yeast of each commercial starter was rehydrated in water at 37 °C for 15 min, and then 7.5 mL of each culture (100 g/L) was used to inoculate individual grape batches.Fermentations were carried out at 22 ± 1 °C, with the cap being broken twice a day by gently pressing it down skins with a steel plunger and density and temperature monitored daily.Upon completion of alcoholic fermentation (glucose-fructose < 2 g/L), wines were pressed, cold settled and then clarified by centrifugation at 4500 rpm for 15 minutes at 5 °C (Avanti J-25, Beckman Coulter, California, USA).Potassium metabisulphite was added to a final free SO 2 concentration of 25 mg/L, after which the wines were bottled in 330 mL glass bottles with crown caps.

Model ageing protocol
An accelerated ageing protocol was employed based on the procedure described by Slaghenaufi et al. (2019).115 mL of wine was placed in glass vials and crimped, leaving mg/L of oxygen.Vials were sealed with epoxy resin to prevent any oxygen ingress, and sample vials were placed at 16 °C (labelled 'young'-T16 ) and 40 °C (labelled 'aged'-T40) for 30 days.

Analysis of volatile compounds
For quantification of terpenes, SPME extraction followed by GC-MS analysis was used, following the procedure described by Slaghenaufi and Ugliano (2018).5 µL of internal standard 2-octanol (4.2 mg/L in ethanol) are added to 5 mL of wine diluted with 5 mL of deionised water in a 20 mL glass vial.3 g of NaCl are added prior to GC-MS analysis.Samples were equilibrated for 1 min at 40 °C.Subsequently, SPME extraction was performed using a 50/30 μm divinylbenzene-carboxen-polydimethylsiloxane (DVB/CAR/PDMS) fibre (Supelco, Bellafonte, PA, USA) exposed to sample headspace for 60 minutes.GC-MS analysis was carried out on an HP 7890A (Agilent Technologies) gas chromatograph coupled to a 5977B quadrupole mass spectrometer, equipped with a Gerstel MPS3 autosampler (Mülheim/Ruhr, Germany).Separation was performed using a DB-WAX UI capillary column (30 m × 0.25, 0.25 μm film thickness, Agilent Technologies) and helium as carrier gas at 1.2 mL/min of constant flow rate.GC oven was programmed as follows: started at 40 °C for 3 min, raised to 230 °C at 4 °C/min and maintained for 20 min.Mass spectrometer operated in electron ionisation (EI) at 70 eV with ion source temperature at 250 °C and quadrupole temperature at 150 °C.Mass spectra were acquired in SIM mode.A calibration curve was prepared for each analyte using seven concentration points and three replicate solutions per point in wine.5 µL of internal standard 2-octanol (4.2 mg/L in ethanol) were added to each calibration point, which was then submitted to SPME extraction and GC-MS analysis as described for the samples.Calibration curves were obtained using Chemstation software (Agilent Technologies, Inc.) by linear regression, plotting the response ratio (analyte peak area divided by internal standard peak area) against concentration ratio (added analyte concentration divided by internal standard concentration).
For quantification of p-menthane-1,8-diol, solid-phase extraction (SPE) followed by GC-MS analysis was used, following the procedure described by Slaghenaufi et al. (2019).100 µL of internal standard 2-octanol (4.2 mg/L in ethanol) are added to samples prepared with 50 mL of wine and diluted with 50 mL of deionised water.Samples were loaded on a BOND ELUT-ENV, SPE cartridge (Agilent Technologies, USA) previously activated with 20 mL of dichloromethane, 20 mL of methanol and equilibrated with 20 mL of water.After sample loading, the cartridges were washed with 15 mL of water.P-Menthane-1,8-diol was eluted with 10 mL of dichloromethane, which was then concentrated under a gentle nitrogen stream to 200 μL prior to GC injection.After elution of p-menthane-1,8-diol, glycosidic precursors were eluted with 20 mL of methanol for enzymatic hydrolysis, which was performed as described by Slaghenaufi et al. (2019).The extracts were transferred in a glass flask and evaporated under vacuum using a Büchi R-215 Rotavapor System (Büchi, Switzerland), redissolved in 5 mL of citrate buffer (pH 5) and added with100 mg of polyvinylpolypyrrolidone (PVPP) and 200 μL of enzyme solution AR2000 (70 mg/mL in citrate buffer).Samples were incubated at 37 °C overnight.Liberated aglycones were extracted as free volatile compounds using the SPE protocol described above.GC-MS analysis was carried out on an HP 7890A (Agilent Technologies) gas chromatograph coupled to a 5977B quadrupole mass spectrometer, equipped with a Gerstel MPS3 autosampler (Mülheim/Ruhr, Germany).Separation was performed using a DB-WAX UI capillary column (30 m × 0.25, 0.25 μm film thickness, Agilent Technologies) and helium as carrier gas at 1.2 mL/min of constant flow rate.GC oven was programmed as follows: started at 40 °C for 3 min, raised to 230 °C at 4 °C/min and maintained for 20 min.Mass spectrometer operated in electron ionisation (EI) at 70 eV with ion source temperature at 250 °C and quadrupole temperature at 150 °C.Mass spectra were acquired in SIM mode.
Calibration curves were prepared using seven concentration points and three replicate solutions per point in model wine (12 % v/v ethanol, 3.5 g/L tartaric acid, pH 3.5) 100 µL of internal standard 2-octanol (4.2 mg/L in ethanol) were added to each calibration solution, which was then submitted to SPE extraction and GC-MS analysis as described for the samples.Calibration curves were obtained using Chemstation software (Agilent Technologies, Inc.) by linear regression, plotting the response ratio (analyte peak area divided by internal standard peak area) against concentration ratio (added analyte concentration divided by internal standard concentration).Retention indices, quantifying and qualifier ions are reported in Supplementary 1.

Statistical analysis
One-way ANOVA and Hierarchical Cluster Analysis (HCA) of chemical data have been performed using XLSTAT 2017 (Addinsoft SARL, Paris, France).Heatmap has been performed with MetaboAnalyst 5.0 (http://www.metaboanalyst.ca,accessed on 16 November 2021) created at the University of Alberta, Edmonton, AB, Canada.

RESULTS AND DISCUSSION
The main oenological parameters of wines are shown in Table 1.Relatively wide ranges were observed for all parameters, reflecting the fact that grapes of different varieties, vintages and geographical origins were employed for winemaking.pH values ranged from 2.77 to 3.31, whereas total acidity from 6.5 to 10 g/L and ethanol from 10.2 to 13.9 % v/v.Acid acetic content was generally below 0.4 g/L.
Terpene content of the young (T16) wines are shown in Table 2 and Supplementary 2-4, indicating rather variable terpene profiles across the sample set.Corvina wines were generally richer than Corvinone.In comparison with data reported in the literature (Black et al., 2015;Slaghenaufi et al., 2022), the contents of the various monoterpene alcohols were relatively high, whereas cyclic and by-cyclic terpenes were generally detected in much lower concentrations.
Glycosylated forms of various monoterpene alcohols were also observed (Table 2 and Supplementary 2-4), the most abundant being linalool, geraniol and nerol.In this case, Corvinone showed higher content compared to Corvina.
Terpene profiles were strongly affected by ageing (Table 2 and Supplementary 2-4), T40 samples), generally showing a decrease in the concentration of monoterpene alcohols such as linalool, β-citronellol and geraniol and an increase in linalool oxides and various cyclic and bi-cyclic terpene and diols.However, as shown in Figure 1a,b, patterns of terpene evolution during ageing were rather complex and appeared to cluster according to different trends.In the case of Corvina, young wines were grouped in two main clusters, one (containing samples W1-5 and W12-15) associated with higher content of linalool, limonene and, in W12-15, also of a-terpineol and g-terpinene, the other (containing samples W6-11) with higher content of geraniol and β-citronellol.Aged samples (T40) were also separated into two clusters, which only in part corresponded to those of young wines.Indeed, one major cluster included W1, 2, 3, 5, 12, 13, 15 and was characterised by a higher content of linalool oxides, cineoles, terpinen 4-ol and p-methane-1,8-diol, while the second cluster (W4 and W6-10) showed moderate levels of cineoles.Moreover, in the case of Corvinone, two clusters were observed for young wines, one (W21-24) associated  with increased content of geraniol and β-citronellol, the other with increased levels of terpinen-4-ol, p-cymene, linalool, and in some wines also of a-terpineol and g-terpinene.
Upon ageing (T40 samples), these clusters were only in part preserved, so that of the wines initially present in the first cluster, only W21-23 were still relatively close, being characterised by moderate to high content of cineoles.
Overall, these observations indicate that, while the terpene profile of aged Valpolicella wines was substantially different from that of the corresponding young wines, the transformations taking place during ageing were not following patterns clearly associated with the initial content of specific 'key' terpenes.Figure 2 summarises the main chemical reactions considered in this study.Free monoterpene alcohols act as potential precursors to a variety of other terpenes through acid-catalysed cyclisations, in particular leading to cyclic terpene alcohols that can, in turn, generate terpene diols and eventually bicyclic terpenes such as cineoles (Skouroumounis and Sefton, 2000;Slaghenaufi and Ugliano, 2018).One exception is the monoterpene alcohol b-citronellol (which contributed to characterising one of the clusters in Figure 1) which is less prone to acid rearrangements or hydration reactions due to the non-allylic structure of the related carbocation.Overall, the fact that monoterpene alcohols were mostly associated with young wines is therefore somewhat expected.At the same time, monoterpene alcohols, particularly geraniol, are also present in glycosidic forms, which can generate free terpenes in the acidic wine environment (Skouroumounis and Sefton, 2000).Likewise, many cyclic terpenes are also present in young wines, as they can originate directly from the grapes or relatively fast reactions occurring during winemaking.
To investigate the relationships between these various groups of terpenes, a correlation study was undertaken (Table 3).Correlation coefficients between free terpenes in T16 and T40 wines were generally low, with the exclusion of linalool and limonene, which were well correlated (R 2 > 0.8) with p-menthane-1,8-diol.This apparent lack of  any clear relationship between the terpene profile of young wines and that of aged wines reflects to some extent the complexity of terpene reactions in wine hydroalcoholic acidic environment as well as suggesting the existence of modulating factors influencing the actual rate of formation/ degradation of the various terpenes.The main terpenes of young wines, in particular linear monoterpene alcohols, are subject to cyclisation and hydration reactions at wine pH, so that their disappearance is somewhat expected, although other terpenes can be generated from them.Some ageingrelated terpenes, including the cyclic monoterpene alcohol α-terpineol, as well as the other cyclic terpenes derived from the degradation of monoterpene alcohols, are intermediates in the process of formation of other cyclic and bicyclic terpenes and, therefore, they might not accumulate in a way that reflects the original content of their possible precursors (Slaghenaufi and Ugliano, 2018).More surprising was the observation that glycosidically-bound forms of the various terpenes detected, which were predominantly of linalool and geraniol, also did not appear to account for any of the terpenes observed in aged wines, with correlation coefficients that were even lower than those of the free compounds.Acid hydrolysis of monoterpene glycosides is considered a significant route to terpene formation at wine pH (Williams et al., 1982), so a certain degree of correlation between glycosylated terpenes of young wines and free terpenes of aged wines would have been expected.However, the formation of terpenes such as cineoles from monoterpene glycosides occurs at extremely harsh pH, while it might be less favoured at wine pH (Williams et al., 1982).Additionally, Skouroumounis and Sefton (2000) observed much higher reaction yields for hydrolytic conversion of free geraniol into linalool, α-terpineol and other cyclic terpenes compared to geraniol glucoside, indicating that, at wine pH, the terpene profile of aged wines could be more dependent of free terpene of young wines.
One set of generally good and significant correlations that were observed across the entire dataset was the one involving linalool, p-menthane-1,8-diol and 1,8-cineole.This was interesting as cineoles have been recently indicated as sensorially relevant to the aroma of aged red wines, where they can contribute to minty, balsamic and hay characters (Antalick et al., 2015;Black et al., 2015;Poitou et al., 2017;Slaghenaufi and Ugliano, 2018).While some authors have reported a possible exogenous origin for cineoles via airborne migration (Hervé et al., 2003), their occurrence in aged wines appears to be mostly related to the degradation of other terpenes during ageing (Fariña et al., 2015, Slaghenaufi andUgliano, 2018).According to Williams et al (1982), cineoles can arise Correlation (R 2 ) between terpenes in aged wines and their free and glycosidically-bound forms in young wines in Corvina and Corvinone wines together.
Free Terpenes T40 Free terpenes Giovanni Luzzini et al. from harsh acid hydrolysis of glycosylated terpenes.Slaghenaufi and Ugliano (2018) proposed at wine pH a formation pathway involving free forms of the monoterpene alcohols linalool, geraniol, nerol and α-terpineol as precursors, and in which p-menthane-1,8-diol is an intermediate in the formation of 1,8-cineole but also 1,4-cineole.We found that p-menthane-1,8-diol in aged samples was very well correlated with the sum of linalool, geraniol, nerol and α-terpineol in young wines (R 2 = 0.7735) and, in particular, with linalool (R 2 = 0.8219), the most abundant monoterpene alcohol in young wines (Figure 3a,b).Conversely, the content of glycosylated terpenes in young wines was not associated with increased p-menthane-1,8-diol formation.
As for cineoles, the sum of free linalool, geraniol, nerol and α-terpineol in young wines was reasonably well correlated (R 2 = 0.6646) with 1,8-cineole in aged wines, whereas the correlation was less strong (R 2 = 0.449) for linalool alone (Figure 4a,b).Considering that the sample set contained wines of three different vintages, which might differ for other matrix-related factors other than terpene profiles, additional insights were obtained when correlations were assessed within individual vintages.Correlations between 1,8-cineole and either the sum of free linalool, geraniol, nerol and α-terpineol or linalool alone were good within each vintage, with vintages 2018 and 2019 showing a similar slope while vintage 2017 showed a lower slope, and thus a lower yield of 1,8-cineole from the two possible precursors (Figure 5a and 5b).Further evaluation of the possible drivers of such differences in the reaction yields indicated that pH was a major factor differentiating the three vintages, with 2017 showing higher and less dispersed values than the other two vintages, which tended to be on the lower end of the pH values.Accordingly, across the three vintages, the wines could be effectively divided into two groups showing different yields, with a pH value of 3 appearing to be the threshold between lower and higher yields (Figure 5c,d).
Overall, a significant negative correlation (p < 0.0001, Pearson) between pH and the yield of 1,8-cineole from either free linalool or the sum of free linalool, geraniol, nerol and α-terpineol was observed (Table 4), so that a decrease in compound yields was observed at increased pH.Regarding 1,4-cineole in aged wines, correlations with either the sum of free monoterpenes as well as of linalool in young wines were not as good as those observed for 1,8-cineole (R 2 = 0.368 and 0.317, respectively).Segmentation of wine samples by either vintage or pH did result in higher correlation coefficients (Figure 6a-d), indicating a certain modulating effect of pH in this case too.However, especially in the low pH samples, several cases were observed in which wines with relatively high total monoterpene or linalool content did not give any 1,4-cineol, as well as one case in which 1,4-cineole levels were much lower than expected.Therefore, other relevant modulation factors are likely to exist, and the relationship between reaction yield and pH appeared less strong for this molecule (Table 4).The formation of 1,4-cineole from monoterpene alcohols requires more steps than that of    1,8-cineole (Slaghenaufi and Ugliano, 2018), so that less robust correlations are somewhat to be expected.Among the additional proposed intermediates leading to 1,4-cineole (Figure 2), the content of the cyclic terpene terpinen-4-ol in aged wines was well correlated with the sum of monoterpene alcohols of young wine (Figure 7), supporting the hypothesis that monoterpene content of young wines contributes to the 1,4-cineole formation during ageing.
Overall, these data indicate that, of the different forms of terpenes present in Valpolicella young red wines, free monoterpene alcohols account mostly for the formation of potent terpene odorants such as cineoles during ageing, Giovanni Luzzini et al.
whereas glycosylated monoterpene alcohols appear to play a minor role.pH is a major modulator of the actual yield of cineoles formation from monoterpene alcohols, so that, depending on pH, young wines that are relatively rich in monoterpenes might not give, upon ageing, high cineoles content, and vice-versa.The influence of pH appears to be related to the last steps of cineoles formation, whereas conversion of the different monoterpene alcohols into the cyclic intermediate p-menthane-1,8-diol was not pH-dependent.

CONCLUSIONS
Ageing can radically change the terpene profile of a wine.The terpene profile of young Valpolicella red wines appeared characterised by a relatively high content of different monoterpene alcohols, in particular in free forms.These free monoterpene alcohols account mostly for the formation of potent terpene odorants such as cineoles during ageing, whereas glycosylated monoterpene alcohols appear to play a minor role.pH is a major modulator of the actual yield of cineoles formation from monoterpene alcohols, so that, depending on pH, young wines that are rich in monoterpenes might not give upon ageing high cineoles content, and vice-versa.The influence of pH appears to be related to the last steps of cineoles formation, whereas conversion of the different monoterpene alcohols into the cyclic intermediate p-menthane-1,8-diol was not pH-dependent.
The results of this study provide novel insights into the mechanism associated with aroma development during ageing, highlighting the significant contribution of the pool of free terpenes in young wines.Further studies should be conducted on model solutions containing different potential precursors and possibly other major matrix components (e.g.phenolics) to validate this observation and further classify red wine aroma ageing potential in other varieties.The observation that the transformation of young wines terpenes into ageing-related ones can be significantly impacted by pH indicate opportunities for fine-tuning aroma development in relation to expected wine shelf-life.

TABLE 1 .
Enological parameters of young wines.
1LOQ of acetic acid is 0.03 g/L.GiovanniLuzzini et al.

TABLE 2 .
Minimum, maximum, mean content, standard deviation of quantified terpenes and p-values (ANOVA) between T16 and T40 wines.