1H NMR metabolite fingerprints of grape berry: Comparison of vintage and soil effects in Bordeaux grapevine growing areas
Introduction
The “terroir” term has been used in the past years to specify the origin of grape and wine production or to express the typicity of a farm-produced product [1]. It takes into account the soil, climate, cultivar, rootstock and viticultural and enological practices [2], [3]. Because the quality of the harvest varies from year to year, vintage has been considered as an important factor for the metabolic composition of grape berries, influencing the sugar, acidity, nitrogen and phenolic compound balance [2]. For grape and wine, vintage is a combination of factors of the regional climate and of the soil. The climate has mainly an effect on the vine phenology by the sum of temperatures, and on the vine water status by the vineyard water balance [4]. Temperature during berry maturation is a major factor that determines grape quality for wine making, which necessitates optimal sugar, organic acid and color content. Climate and soil interact for the vine phenology and vine water and mineral status. They are linked through the soil water content, the main characteristic describing a vineyard [3]. In a geographical region, where many local Appellation of Controlled Origin (ACO) wines are identified, the hierarchy between the vintage effects and the soil effects on grape quality is still under debate.
Proton nuclear magnetic resonance (1H NMR) spectroscopy is a powerful tool to determine metabolite fingerprinting and/or metabolic profiling of plant or animal extracts [5], [6], [7]. 1H NMR allows the simultaneous detection of proton-containing compounds in a complex mixture, and is used to determine the cell or tissue extract composition without a previous knowledge of the samples [7], [8]. 1H NMR spectroscopy associated to multivariate statistical analysis has been used to study effect of biotic [9] or abiotic stress [10] on composition changes of numerous plants as well as to detect potential unintended effects following a genetic modification [11], [12]. In food science, such techniques have also been found suitable for metabolite profiling in fruit and fruit juices [13], [14], potato [11], olive oil [15], tea [16], beer [17] and wine [18].
Chemometric methods, as principal component analysis (PCA) and partial least squares (PLS) [19], allow to describe sample clustering and to detect the biochemical compounds responsible for the separation of samples. PCA is essentially a descriptive method used to visualize samples present in an n-dimensional space of a starting set of variables into a smaller number of dimensions, called principal components (PCs), that represent sources of successively maximized variance of data [20], [21]. PLS regression, like PCA, identifies synthetic variables (scores) that describe the variance in a sample set. But PLS uses additional information: the a priori definition of the sample groups [19]. The scores are optimized to separate the groups. Another output is to reveal the most effective variables that allow the groups to be separated. After PLS calibration, a model can be set with the most effective variables in order to classify unknown samples [19], [22], [23]. The use of chemometric methods to discriminate metabolic profiles allows the detection of unexpected or unpredicted changes to characterize different genotypes and phenotypes [24].
Recently, the combination of 1H NMR spectroscopy with PCA has been used to describe the variability in the composition of skin and pulp tissues of grape berries harvested at maturity in four appellations in the Bordeaux area in 2002 [25]. A good clustering of the metabolic profiles of pulps or skins in relation to their appellation was observed for this vintage. The 1H NMR results showed sample clustering similar to those determined using the components routinely measured to determine the potential quality of grapes for winemaking (sugars, acidity, pH and total phenolics). Therefore, the combination of 1H NMR spectroscopy with multivariate statistical analysis was able to discriminate berry samples from different appellations of the same vintage. The question remains whether this approach is still valid for samples of different cultivars harvested on different vintages.
The aim of the present work was to study the variability of metabolite fingerprints of grape berry skin extracts using 1H NMR spectra. These fingerprints were obtained from the three major red cultivars used in Bordeaux vineyards (‘Merlot noir’, ‘Carbernet franc’ or ‘Cabernet Sauvignon’), growing in five small wine production locations in the Aquitaine Region (Buzet, Pessac-Léognan, Saint-Emilion, Médoc and Pomerol) on three vintages (2002, 2003 and 2004). Chemometric analyses (PCA and PLS) were applied on the 1H NMR data to visualize clustering of samples and to detect the metabolites responsible for the discrimination of sample groups. The results reported here show that the vintage factor is much more discriminant on berry quality analyzed by 1H NMR metabolite fingerprinting than the soil factor. The major compounds identified in the 1H NMR spectra and responsible for the discrimination between groups included soluble sugars, amino and organic acids.
Section snippets
Origin of the samples
Numerous samples were required to make a robust evaluation of the method. A total of 277 samples of ripe grape berries of ‘Merlot noir’, ‘Carbernet franc’ or ‘Cabernet Sauvignon’ cultivars were harvested on three vintages, 2002, 2003 and 2004. Five different vineyards areas (located in five ACOs) in Bordeaux-France were studied (Buzet, Pessac-Léognan, Saint-Emilion, Médoc and Pomerol). The plots were chosen in different conditions of soil, cultivars and vintages to represent the variability of
Multivariate statistical analysis of 1H NMR data from ‘Merlot noir’, ‘Carbernet franc’ and ‘Cabernet Sauvignon’ cultivars according to three vintages
Two hundred and seventy-seven samples of grape berries of ‘Merlot noir’ (MN), ‘Carbernet franc’ (CF) or ‘Cabernet Sauvignon’ (CS) cultivars were harvested from five wine-growing areas in Bordeaux during 2002, 2003 and 2004 vintages. A typical 1H NMR spectrum of berry skin is presented Fig. 1. PCA analysis on 183 spectral domains of each 1H NMR spectrum for each skin extract was used to explore the variability of the cultivar, terroir and vintage factors. The cultivar factor was not discriminant
Conclusions
When discriminant analysis was applied on 1H NMR spectra of water-soluble metabolites of mature berry skin, PLS analysis produced a set of synthetic variables that separated clearly samples of berries grown in a large geographical area in the Bordeaux region. After a linear discriminant analysis, a model allowed to sort test samples very efficiently. 1H NMR seems an advantageous technique to characterize berries grown under different climates. The first PLS score was described by phenolics on
Acknowledgements
We thank The National Council to Scientific and Technologic Development (CNPq-Brazil) for a grant (G.E. Pereira), the “Conseil Interprofessionnel des Vins de Bordeaux” and “Aquitaine Region” (France) for their financial support to the project, and the wineries that provided the grape berry samples. NMR analyses were done on the Metabolome-Fluxome Facility of Bordeaux (http://www.bordeaux.inra.fr/umr619/UK_page_PF_metabolome.htm).
References (34)
- et al.
Anal. Chim. Acta
(2004) Progr. Nucl. Magn. Reson. Spectrosc.
(1996)- et al.
Phytochemistry
(2003) - et al.
Food Chem.
(1998) - et al.
Anal. Chim. Acta
(2002) Chemom. Intelligent Lab. Sys.
(1996)- et al.
Phytochemistry
(2003) - et al.
J. Food Eng.
(2003) J. Proc. Cont.
(2004)- et al.
Anal. Chim. Acta
(2003)
Food Chem.
J. Food Eng.
Les Terroirs Viticoles
Amer. J. Enol. Vitic.
Bull. OIV
Funct. Plant Biol.
Funct. Plant Biol.
Cited by (151)
Emerging studies of NMR-based metabolomics of fruits regarding botanic family species associated with postharvest quality
2024, Journal of Food Composition and AnalysisUntargeted metabolomics analysis based on LC-IM-QTOF-MS for discriminating geographical origin and vintage of Chinese red wine
2023, Food Research InternationalTerroir: The effect of the physical environment on vine growth, grape ripening, and wine sensory attributes
2021, Managing Wine Quality: Volume One: Viticulture and Wine Quality