Oak aging is a crucial step in winemaking, during which the organoleptic properties of wine are modified. Various parameters affect the chemical composition of oak, and botanical origin previously has been shown to be a determinant factor.

Sweet natural triterpenes arising from cooperage oak and the effect of oak species (Quercus petraea and Quercus robur) on four recently discovered taste-active triterpenes (three sweet and one bitter) were analyzed. The results showed that sessile oak or Quercus petraea (27 samples) was richer in sweet triterpenes and poorer in the bitter one than pedunculate oak or Quercus robur (19 samples), with high inter-individual variations within species.

A triterpenoid index was calculated to reveal the triterpenoid composition of oak. This index appeared as a promising tool for the unambiguous discrimination of oak species and could offer new insights concerning oak selection by coopers and the monitoring of élevage by winemakers. 

The quality of a wine relies both on the value of the harvested grapes and control of its winemaking process. For many wines, a maturation period involves contact between wine and oak wood, wine bouquet and taste being profoundly modified during this step. These modifications can be due to the moderate oxidation of wine compounds during élevage or to the release of molecules from wood.¹

The research in this field has led to identification of the key volatile compounds coming from oak: vanillin, β-methyl-γ-octalactone (oak-lactone), eugenol and 2-furanmethanethiol.^2,3 To a large extent, these volatiles explain the vanilla, coconut, spicy and roast coffee aromas typical of oaked wines.

Moreover, oak releases non-volatile compounds likely to modify the taste of wine. In practice, winemakers observe the modification of tannin perception of wine (structure and dryness) and bitterness as well as perceptions described as volume or sweetness.

Much research has focused on ellagitannins, a specific class of wood tannins. The sensory properties (bitterness and astringency) of isolated ellagitannins have been investigated, and recent research has described the determination of their perception threshold, thanks to a half-tongue test.^4-6

Products of interaction between ellagitannins and grape flavonoids also were identified in red wine; they could be involved in the color change of wine during élevage.⁷ Beyond ellagitannins, other non-volatile compounds are released from oak such as coumarins⁸ and lignans.^9-11 However, little research data has referred to the sensory properties of non-volatile compounds and more particularly to sweet components of the taste.

Various parameters such as the origin of oak or cooperage techniques can influence the composition of wood and subsequently its effect on wine taste. Some coopers classify the qualities of wood according to ring width (also called grain) or geographic origin (forests). However, previous studies have demonstrated that species is a better indicator of chemical composition than morphological parameters or provenance.^12,13

In practice, two main species of European oak are used for cooperage and occur together in most French forests: Quercus petraea Liebl (sessile oak) and Quercus roburL. (pedunculate oak). As reported by various authors, the average levels in oak-lactone and ellagitannins are respectively higher and lower in sessile oak than in pedunculate oak.^13-16 But high inter-individual variations are observed within each species and deeply affect the significance of the species effect.^17-19

For example, A. Prida et al. showed that some sessile oaks have low levels of oak-lactone in the range of those assayed in pedunculate oaks.¹⁸ A similar situation was described for ellagitannins.^13,20 The odorant and taste-active compounds seem not specific to either sessile or pedunculate oak, and the quantification of oak lactone or ellagitannins in wood samples does not allow discrimination between the oak species.

In this research, the authors present data related to the identification of a new class of sweet wood compounds and distribution of these compounds in Quercus roburand Quercus petraea. This data enabled the establishment of a method to accurately discriminate between the aforementioned species.

Identification of sweet triterpenosides occurrence in wine
The novel methodology of taste-active compound identification was developed and implemented in this research. The method consists first in off-line centrifugal partition chromatography allowing the researchers to obtain 15 different wood fractions, according to their affinity to different solvents used in the study (n-heptane/ethyl acetate/methanol/water).

The collected fractions were freeze-dried and tasted in order to find ones characterized by sweet taste. The most prominent ones were analyzed and then purified using preparative HPLC. Finally, purified compounds were analyzed using Fourier transform mass spectrometry (FT-MS, Orbitrap analyzer) jointly with two-dimensional nuclear magnetic resonance (2D ¹H and ¹³C NMR). This technique allowed obtaining the structural elucidation of the purified compounds. The tandem mass spectrometry (MS/MS) spectra obtained with resonant and non-resonant fragmentation modes were compared, thus providing complementary information about the molecular structure.

Two oleanane-type triterpenoids substituted with galloyl and glucosyl moieties were identified for the first time, one of which exhibits sweet properties. These compounds, which have never been reported, were named Quercotriterpenoside (QTT) I and II.²¹

Sensory threshold and occurrence of wine
The purified compounds were characterized by a strong, sweet taste. They were tasted in white wine, and the sensory thresholds were measured using International Organization for Standardization (ISO) procedure. The threshold found was 590 µg/L for QTT I, which made this compound a very prominent marker of sweet taste arising from élevage. This sensory threshold is the lowest among compounds that could be observed in wine. The analysis of wines aged in oak barrels demonstrated that the amount of QTT in wines varies widely and can reach 1,000 µg/L for the sum of QTT. This fact proves the implication of this compound in sweet perception at least for one part of cited wines.

Other triperpenosides
Further research identified a group of similar compounds coming from the same family, differing by galloyl and glycoside groups as well as their isomers.²²

Sensory properties of these compounds were not completely elucidated, but they are more likely characterized by the sweet taste. Their concomitant occurrence in oak lets us think about cumulative impact on the wine sweetness and the possibility of modulating of taste balance.

The group of other triterpenoids carrying carboxylic acid function on carbon 24 of the non-sugar part of molecules and more particularly glucosyl derivative of bartogenic acid (Glu-BA) attracted our attention. This compound (see "Chemical Structures of Quercotriterpenosides Isomers and Glucosyl Derivative of Bartogenic Acid"), first identified by G. Arramon et al.,⁹exhibits a bitter taste and had very specific distribution in oak extracts in comparison with occurrence of the QTT family.

Impact of botanical species on QTTs and Glu-BA
There are two major European oak species used in cooperage (Quercus robur and Quercus petraea). Previous research suggested that the difference between species arises from geographic origin. But in some cases this difference is not constant regarding known chemical markers.

Current research deals with analysis of newly identified QTT and Glu-BA in wood coming from both species.

Forty-six samples of fresh oak material were collected from eight French forests: three in northeast (Saint-Clément, Spincourt and Xures), two in center (Tronçais and Chateauroux), one in northwest (Liffré), one in southwest (Pierroton) and one in southeast (Laveyron). Samples from Pierroton and Laveyron were provided by the French National Institute for Agricultural Research and Dr. Erwan Guichoux. The other samples were supplied by Seguin-Moreau cooperage. The distribution of the samples is provided in "French Oak Sourcing Origins" above.

Each sample represented oak shavings from one single living oak tree drilled at a height of 1 meter. The shavings were collected in a plastic bag filled with silica gel beans. Green leaves from the same tree were collected in a separate plastic bag filled in the same way with silica gel beans. Both samples from one tree were identified with the same code and sent to the laboratory. A genetic analysis was run on leaves immediately after sample reception.

Genetic assignment of species
The genetic assignment method was used in this study for the oak samples. It consists of assigning individual trees to species (Quercus robur or Quercus petraea) based on the expected frequencies of their genotypes, as described by E. Guichoux et al. ^23,24

The percentage of assignment shows the probability of a tested individual tree to belong to one species rather than another. The statistical threshold is 87.5%, which allows assigning individual trees to a single species. The individuals being assigned with probability higher than 87.5% to one specific species are either Q. robur(pedunculate oak) or Q. petraea (sessile oak), while individuals assigned with probability lower than 87.5% should be considered hybrids (hybrid F1, F2, backcross of first generation).

Chemical analysis: Oak wood shavings were dried, ground into a homogeneous powder, soaked in model solution and analyzed by U-HPLC coupled with Fourier Transform Mass Spectrometry (FTMS).²⁵ Four compounds were quantified in oak extracts: QTT I, QTT II, QTT III and Glu-BA. The quantification method was validated by studying sensitivity, linearity in working range, intraday repeatability, interday precision, trueness and specificity.

Genetic discrimination of oak samples: Fresh leaves were used to facilitate DNA extraction. Among 46 samples analyzed in this study, 27 were assigned to Q. petraea (sessile oak) and 19 to Q. robur(pedunculate oak). Both species were found in some forests (Tronçais and Liffré), confirming that geographic origin is not in itself a relevant element to discriminate sessile and pedunculate oak.

Triterpenoids determination in oak wood samples: The method developed in this study was applied to quantify for the first time QTT I, II and III in oak. The glucosyl derivative of bartogenic acid (Glu-BA) was concomitantly measured. The four considered triterpenoids were observed in all samples, and all the concentrations were measured above their limit of quantification.

For each compound, the results showed a wide range of concentration with more than four orders of magnitude (0.7-1,102.5 µg/g for QTT I; 0.7-1,418.5 µg/g for QTT II; 1.2-1,408.7 µg/g for QTT III and 3.2-1,952.0 µg/g for Glu-BA). Samples with high content in QTT I seemed, to a large extent, also rich in QTT II and QTT III and reciprocally. Pearson tests were realized on the 46 samples and exhibited correlations between QTT I and QTT II (R² = 0.76), QTT I and QTT III (R² = 0.74), QTT II and QTT III (R² = 0.77) with p-values less than 0.1% for each pair. All these correlations were positive, suggesting a similar biosynthetic pathway for these compounds.

In contrast, the overall relation between Glu-BA and QTTs concentration appeared more complex. Pearson's test was significant (p-value less than 0.1%), but with a weak negative correlation (R² = 0.25 with QTT I; R² = 0.13 with QTT II, and R² = 0.20 with QTT III) sessile oak samples appeared as being rather rich in QTT I and rather poor in Glu-BA, whereas pedunculate oak samples seemed to be rather poor in QTT I and rather rich in Glu-BA. No compound was species-specific: All triterpenoids were present in all samples.

To observe more precisely the relationship between the triterpenoids composition and the botanical species, the mean concentration of the quantified compounds were calculated for sessile (n = 27) and pedunculate (n = 19) oak samples (see "Mean Concentrations of QTT I, QTT II, QTT III and Glu-BA in Sessile and Pedunculate Oak").

For QTT I, the mean values were 413.5 ± 96.2 µg/g for sessile oak samples and 6.0 ± 2.7 µg/g for pedunculate oak. Similar results were obtained for QTT II and QTT III, demonstrating that sessile oak was richer in QTTs than pedunculate oak.

The Glu-BA mean concentration was higher in pedunculate oak (795.3 ± 271.3 µg/g) than in sessile oak (24.4 ± 10.7 µg/g).

Application of one-way analysis of variance revealed significant differences between species for all compounds (p-value less than 0.1 %). This trend was similar for samples of different species coming from the same forest, suggesting that the botanical species had a predominant influence on triterpenoids composition of oak in comparison with geographic location.

These results could be of particular interest regarding the organoleptic effect of oak aging on wine taste. Indeed, QTT I, II and III develop a sweet taste, whereas Glu-BA has been described as bitter. The present study highlighted that sessile oak contained more sweet triterpenoids, whereas pedunculate oak was richer in bitter triterpene.

Although statistical tests revealed significant differences for mean concentration in QTTs and Glu-BA, some extreme values of individual triterpenes were very close for sessile and pedunculate oak (see "Mean, Minimal and Maximum Values Observed in Sessile and Pedunculate Oak"). For example, the minimal concentration in QTT II measured for sessile oak samples was 23.6 µg/g, whereas the maximum value for pedunculate oak samples was 44.1 µg/g. For Glu-BA, the maximum concentration of sessile samples was 105.5 µg/g, and the minimum concentration was 36.0 µg/g for pedunculate samples.

Mean amounts of QTTs and Glu-BA were respectively higher and lower in sessile oak than in pedunculate oak. High inter-individual variations were observed within species for each triterpenoid, as reflected by large confidence intervals. As a consequence, the individual quantification of each triterpenoid did not allow direct identification of the botanical species. This limitation can be linked with observations concerning other compounds whose concentrations depend on botanical species.

A significant number of sessile oak samples contained levels of β-methyl-γ-octalactone similar or even lower than pedunculate oak samples.¹⁸ A similar trend was observed for ellagitannins.²⁰ None of these compounds (oak-lactone, ellagitannins or triterpenoids) allow an unambiguous discrimination of oak species, according to their individual concentration in wood.

Differentiation according to a triterpenoids index
Beyond absolute concentrations in individual triterpenoids, it seemed that samples could be grouped in two categories according to their relative amounts of QTTs and Glu-BA. To express this relative composition, a triterpenoids index (TI) was calculated as base 10 logarithm of the ratio between the sum of concentrations in QTTs and the concentration in Glu-BA (all given in µg/g).

 

TI = log  [QTT I] + [QTT II] + [QTT III]
                          [Glu - BA]

The average values of this index were calculated for sessile and pedunculate oak wood samples (see "Mean Logarithm of the Ratio Between "QTT and Glu-BA" and "Mean, Minimal and Maximum Values of Observed in Sessile and Pedunculate Oak").

Mean TI was positive for sessile samples and negative for pedunculate samples (1.9 and -1.5 respectively). The application of one-way analysis of variance test revealed that these differences were statistically significant (p-value less than 0.1%). More interesting, confidence intervals were smaller than for absolute concentrations in triterpenes expressing a narrower range of values. All samples of sessile oak had positive TI values (1.2 - 2.4), whereas all samples of pedunculate oak exhibited negative TI values (-2.2 to -0.8).

Contrary to absolute concentrations in individual triterpenoids, there was a huge gap (two log points) between the closest values of both species, i.e., between the lowest value of sessile oak (1.2) and the highest value of pedunculate oak (-0.8).

Consequently, the calculation of a TI reflecting the relative composition in triterpenoids of wood appeared to avoid any ambiguity in the assignment of the botanical species. In practice, a positive TI value might indicate that the sample was from sessile oak, whereas a negative TI value might correspond to a pedunculate oak sample. These results highlighted probable differences between species in terms of triterpenoids biosynthesis, with variations of the enzymes involved in the decoration of the non-sugar parts of molecules. Such differences already had been observed and related to the great diversity of triterpenoids observed in the plant kingdom. ^{26, 27}

Identification of natural sweet compounds
This research confirmed empirical knowledge about imparting sweetness with barrel élevage.

The application of centrifugal partition chromatography coupled with gustotemetry allowed the identification of a class of natural wood compounds characterized by a strong, sweet taste. These compounds could be considered an important contributor to the taste balance of barrel-aged wines.

The analysis of sweet (QTTs) and bitter (Glu-BA) triterpenoids in oak wood provide new insights into the chemical composition of oak wood and interpretation of its organoleptic effect on wine and spirits. This study showed that sessile oak was rich in sweet QTTs and poor in bitter Glu-BA, while pedunculate oak samples exhibited high levels of Glu-BA and low levels of QTTs. This distinction can be related to the preferential use of sessile oak for élevage.

The practical importance of this finding consists in allowing for cooper and winemaker to adapt its wood selection for making barrels with a "sweet" oak profile-or, in contrast, to decrease the "sweet" component if needed.

Analytical criterion of species differentiation
Beyond this significant trend, the large inter-individual variations of concentrations observed for these compounds within each species was similar to previous observations concerning oak-lactone or ellagitannins. Individually, none of these compounds was sufficient to discriminate sessile and pedunculate oak wood. However, a Triterpenoids Index (TI) reflecting the triterpenoids profile of wood samples was calculated and showed a clear-cut differentiation between species without any equivocal sample or recovery between species. The measurement of this TI appeared consequently as a promising tool to identify oak species by a chemical method in addition to genetic assignment.

Until now, unambiguous chemical differentiation of oak species had only been achieved by use of non-targeted analysis with FT-ICR followed by statistical treatment.²⁸ Such an approach gives an accurate fingerprint of the studied sample, and its species assignment is based on the comparison of this fingerprint to a database collection of previously analyzed samples.

Moreover, as FT-ICR is a very powerful but highly costly technique, its use is reserved for few laboratories specialized in mass spectrometry. This constitutes a limitation for the routine application of this technique by the cooperage industry. In contrast, the targeted method presented here involves the LC-MS quantification of only four compounds with straightforward sample preparation and short analysis time. The species assignment can be obtained directly, without any statistical treatment or comparison with a database.

This new, discriminating method was based on taste-active molecules, likely to modify the organoleptic properties of wine. It appears promising for various applications, in particular for a better selection of oak used in cooperage, allowing a more harmonious marriage with wine.

 

References
1. Ribéreau-Gayon, P., Y. Glories, A. Maujean and D. Dubourdieu. 2006 Handbook of Enology. Vol. 2. The Chemistry of Wine Stabilization and Treatments.  Vol. 2.
2. Chatonnet, P., D. Dubourdieu and J.-N. Boidron. 1991 “Effects of fermentation and maturation in oak barrels on the composition and quality of white wines.” Aust. NZ Wine Ind. J.  6, 73–84.
3. Tominaga, T., L. Blanchard, P. Darriet and D. Dubourdieu. 2000 “A powerful aromatic volatile thiol, 2-furanmethanethiol, exhibiting roast coffee aroma in wines made from several Vitis vinifera grape varieties. J. Agric. Food Chem. 48 (5), 1799–1802.
4. Glabasnia, A. and T. Hofmann. 2007 “Identification and sensory evaluation of dehydro- and deoxy-ellagitannins formed upon toasting of oak wood (Quercus alba L.).” J. Agric. Food Chem.  55 (10), 4109–4118.
5. Glabasnia, A. and T. Hofmann. 2006 “Sensory-directed identification of taste-active ellagitannins in American (Quercus alba L.) and European oak wood (Quercus robur L.) and quantitative analysis in bourbon whiskey and oak-matured red wines.” J. Agric. Food Chem. 54 (9), 3380–3390.
6. Stark, T., N. Wollmann, K. Wenker, S. Lösch, A. Glabasnia and T. Hofmann. 2010 “T. Matrix-calibrated LC-MS/MS quantitation and sensory evaluation of oak ellagitannins and their transformation products in red wines. J. Agric. Food Chem. 58 (10), 6360–6369.
7. Chassaing, S., D. Lefeuvre, R. Jacquet, M. Jourdes, L. Ducasse, S. Galland. A. Grelard. C. Saucier, P.-L. Teissedre and O. Dangles et al. 2010 “Physicochemical Studies of New Anthocyano-Ellagitannin Hybrid Pigments: About the Origin of the Influence of Oak C-Glycosidic Ellagitannins on Wine Color.” Eur. J. Org. Chem. (1), 55–63.
8. Moutounet, M., P.H. Rabier, J.L. Puech, E. Verette and J.M. Barillere. 1989 “Analysis by HPLC of extractable substances in oak wood. Application to a Chardonnay wine.” Sci. Aliments (1), 35–51.
9. Arramon, G., C. Saucier, D. Colombani and Y. Glories. 2002 “Identification of triterpene saponins in Quercus robur L. Q. petraea Liebl. heartwood by LC-ESI/MS and NMR.” Phytochem. Anal. 13 (6), 305–310.
10. Marchal, A., B.N. Cretin, L. Sindt, P. Waffo-Téguo and D. Dubourdieu. 2015 “Contribution of oak lignans to wine taste: Chemical identification, sensory characterization and quantification.” Tetrahedron, 71 (20), 3148–3156.
11. Cretin, B.N., Q. Sallembien, L. Sindt, N. Daugey, T. Buffeteau, P. Waffo-Teguo, D. Dubourdieu and A. Marchal. 2015 “How stereochemistry influences the taste of wine: Isolation, characterization and sensory evaluation of lyoniresinol stereoisomers.” Anal. Chim. Acta 888, 191–198.
12. Feuillat, F., L. Moio, E. Guichard, M. Marinov, N. Fournier and J.-L. Puech. 1997 “Variation in the concentration of ellagitannins and cis- and trans-β-methyl-γ-octalactone extracted from oak wood (Quercus robur L., Quercus petraea Liebl.) under model wine cask conditions.” Am. J. Enol. Vitic. 48 (4), 509–515.
13. Doussot, F., P. Pardon, J. Dedier and B. De Jeso. 2000 “Individual, species and geographic origin influence on cooperage oak extractible content (Quercus robur L. and Quercus petraea Liebl.).” Analusis 28 (10), 960–965.
14. Chatonnet, P. and D. Dubourdieu. 1998 “Comparative study of the characteristics of American white oak (Quercus alba) and European oak (Quercus petraea and Q. robur) for production of barrels used in barrel aging of wines.” Am. J. Enol. Vitic. 49 (1), 79–85.
15. Mosedale, J.R. and P.S. Savill. 1996 “Variation of heartwood phenolics and oak lactones between the species and phenological types of Quercus petraea and Q. robur.” Forestry 69 (1), 47–55.
16. Guchu, E., M.C. Díaz-Maroto, I.J. Díaz-Maroto, P. Vila-Lameiro and M.S. Pérez-Coello. 2006 “Influence of the species and geographical location on volatile composition of spanish oak wood (Quercus petraea Liebl. and Quercus robur L.).” J. Agric. Food Chem. 54 (8), 3062–3066.
17. Masson, G., M. Moutounet and J.L. Puech. 1995 “Ellagitannin Content of Oak Wood as a Function of Species and of Sampling Position in the Tree.” Am. J. Enol. Vitic. 46 (2), 262–268.
18. Prida, A., A. Ducousso, R.J. Petit, G. Nepveu and J.L. Puech. 2007 “Variation in wood volatile compounds in a mixed oak stand: Strong species and spatial differentiation in whisky-lactone content.” Ann. For. Sci. 64 (3), 313–320.
19. Prida, A. and J.L. Puech. 2006 “Influence of geographical origin and botanical species on the content of extractives in American, French, and East European oak woods.” J. Agric. Food Chem. 54 (21), 8115–8126.
20. Prida, A., J.C. Boulet, A. Ducousso, G. Nepveu and J.L. Puech. 2006 “Effect of species and ecological conditions on ellagitannin content in oak wood from an even-aged and mixed stand of Quercus robur L. and Quercus petraea Liebl.” Ann. For. Sci.  63 (4), 415–424.
21. Marchal, A., P. Waffo-Téguo, E. Génin, J.M. Mérillon and D. Dubourdieu. 2011 “Identification of new natural sweet compounds in wine using centrifugal partition chromatography-gustatometry and Fourier transform mass spectrometry.” Anal. Chem. 83 (24), 9629–9637.
22. Marchal, A., E. Génin, P. Waffo-Téguo, A. Bibès, G. Da Costa, J.M.  Mérillon and D. Dubourdieu.a, b 2015 “Development of an analytical methodology using Fourier transform mass spectrometry to discover new structural analogs of wine natural sweeteners.”  Analytica Chimica Acta, 853 (1), 425–434
23. Guichoux, E., L. Lagache, S. Wagner, P. Léger and R.J. Petit. 2011 “Two highly validated multiplexes (12-plex and 8-plex) for species delimitation and parentage analysis in oaks (Quercus spp.).” Mol. Ecol. Resour.  11 (3), 578–585.
24. Guichoux, E., P. Garnier-Géré, L. Lagache, T. Lang, C. Boury and R.J. Petit. 2013 “Outlier loci highlight the direction of introgression in oaks.” Mol. Ecol. 22 (2), 450–462.
25. Marchal, A., A. Prida and D. Dubourdieu. 2016 “New Approach for Differentiating Sessile and Pedunculate Oak: Development of a LC-HRMS Method To Quantitate Triterpenoids in Wood.” J.  Ag. & Food Chemistry. 64 (3): 618-626
26. Augustin, J.M., V. Kuzina, S.B. Andersen and S. Bak. 2011 “Molecular activities, biosynthesis and evolution of triterpenoid saponins.” Phytochemistry 72 (6), 435–457.
27. Vincken, J.P., L.  Heng, A. de Groot and H. Gruppen. 2007 “Saponins, classification and occurrence in the plant kingdom.” Phytochemistry 68 (3), 275–297
28. Gougeon, R.D. M. Lucio, A. De Boel, M. Frommberger, N. Hertkorn, D. Peyron, D. Chassagne, F. Feuillat, P. Cayot and A. Voilley et al. 2009 “Expressing forest origins in the chemical composition of cooperage oak woods and corresponding wines by using FTICR-MS.” Chem. - Eur. J.  15 (3), 600–611.