Abstract
The effects of wine yeasts and malolactic bacteria on the anthocyanin profile of Sangiovese wines produced by both commercial- and laboratory-scale winemaking processes were investigated. The mean anthocyanin profile of the wines obtained from commercial winemaking processes, carried out in several wineries in Tuscany, showed a distinctive anthocyanin pattern characterized by high percentages of malvidin-3-glucoside, without appreciable levels of acylated anthocyanins. This anthocyanin pattern, also shared by all wines produced in the laboratory, reproduces the peculiar profile of Sangiovese grapes and distinguishes Sangiovese wines from Merlot and Cabernet Sauvignon wines. Principal component analysis of all anthocyanin data showed that the Sangiovese wines did not group on the basis of wine-producing area, harvest year, or winery but were distributed along a longitudinal axis, owning to the variability of cyanidin-3-glucoside, peonidin-3-glucoside, and malvidin-3-glucoside percentages. This variability was independent of the yeast ecology during fermentation, at least if the yeast ecology was described at the level of yeast species. In both commercial and experimental wines, higher percentages of vitisin A, a malvidin-3-glucoside derivative, were dependent on the growth and dominance of Candida zemplinina in the early stages of vinification. The anthocyanin profile of the Sangiovese wines was also maintained after malolactic fermentation.
- wine anthocyanins
- Sangiovese
- Saccharomyces cerevisiae
- Kloeckera apiculata
- Candida zemplinina
- Oenococcus oeni
The color of red wines, an important quality factor, is primarily dependent on anthocyanins (anthocyanidin-glycosides), phenolic secondary metabolites that accumulate in the grape skin. Vitis vinifera grapes usually contain five glycosylated anthocyanidins: delphinidin-3-glucoside, cyanidin-3-glucoside, petunidin-3-glucoside, peonidin-3-glucoside, and malvidin-3-glucoside, which can be acylated on the glucose moiety to form, as major acylated glucosides, acetyl and p-coumaroyl derivatives. The concentration of these compounds in grape berries is known to vary with grapevine variety and is influenced by viticultural and environmental factors (Downey et al. 2004, 2006, Jackson 2000, Revilla et al. 2001, Yokotsuka and Singleton 1997). However, although anthocyanin concentration in grape skin is highly variable, it is commonly accepted that the anthocyanin profile of a given cultivar, determined by the relative proportions among the different anthocyanins occurring in the berry, is closely linked to its genetic inheritance and is independent of seasonal conditions or producing area. Therefore, the anthocyanin profile and the ratio between individual anthocyanins have been suggested as a fingerprint for chemotaxonomic classification of grape varieties (Ortega-Regules et al. 2006). Moreover, results from anthocyanin analysis could be used to assess red wine authenticity, at least if grapes from which the wine has been obtained have a distinctive feature, as might occur in wines made from Sangiovese grapes.
Sangiovese is the most cultivated red-berried grapevine in the Tuscany region (Italy) and is used as an obligatory variety in the production of some well-known wines such as Chianti Classico and Brunello di Montalcino. According to the current disciplinary regulations, these certified (DOCG, Denominazione di Origine Controllata e Garantita) wines must be produced with a minimum ratio of 80% of Sangiovese grapes for Chianti Classico and exclusively with Sangiovese grapes for Brunello di Montalcino. Sangiovese grapes are known to have a distinctive anthocyanin pattern, characterized by the prevalence of malvidin-3-glucoside and a very low percentage of acylated anthocyanins (Mattivi et al. 2006, Ramazzotti et al. 2008). This latter feature also seems to characterize the wines made from Sangiovese (Castellari et al. 1998, 2001), but there are few studies on the anthocyanin profile of these wines. Indeed, during winemaking, the anthocyanin profile of the grapes is known to undergo some modifications because of physicochemical and biological factors. In this connection, it is well known that the transformation of grape juice into wine is a complex microbial reaction involving the sequential development of various species of yeasts, responsible for alcoholic fermentation (AF), and of lactic acid bacteria (LAB), responsible for malolactic fermentation (MLF). These microbial populations could potentially affect the anthocyanin profile of a red wine through three distinct processes: anthocyanin adsorption on cell walls, microbial metabolite-mediated formation of anthocyanin derivatives, and anthocyanin hydrolysis.
Most research on microbial anthocyanin adsorption has focused on yeast parietal adsorption (Caridi 2007), characterized by weak and reversible interactions between anthocyanins and cell walls (Mazauric et al. 2005). Anthocyanin adsorption is considered a strain-dependent property within Saccharomyces species, variable with the polarity of the anthocyanins (Morata et al. 2005, Vasserot et al. 1997), probably because it is affected by the variable cell wall composition, specifically polar groups exposed on the surface (Morata et al. 2005). Microbial metabolite-mediated formation of anthocyanin derivatives should distinguish among yeast and LAB activities. Yeasts, during AF, release secondary metabolic products such as pyruvic acid and acetaldehyde, which can react with anthocyanins to produce more stable colored derivatives such as vitisin A, vitisin B, and ethyl-linked anthocyanin-flavanol pigments (Medina et al. 2005, Morata et al. 2003, 2007). Lactic acid bacteria, conversely, can consume pyruvic acid and acetaldehyde and thereby limit the production of these anthocyanin-derived pigments (Asenstorfer et al. 2003, Osborne et al. 2000). Finally, recent works have also demonstrated that wine microorganisms could produce extracellular glycosidases and therefore have the potential to hydrolyze anthocyanidin-glycoconjugates, affecting the wine anthocyanin profile (Grimaldi et al. 2000, Liu 2002, McMahon et al. 1999, Spagna et al. 2002).
The aim of this work was to investigate on the effects of both Saccharomyces and non-Saccharomyces wine yeasts and malolactic bacteria on the anthocyanin profile of Sangiovese wines as produced by both commercial- and laboratory-scale winemaking processes.
Materials and Methods
Commercial winemaking processes.
Several winemaking processes, carried out in different wineries in Tuscany (Italy), were investigated by determining the yeast and LAB populations dominating AF and MLF, respectively, and the anthocyanin profile of the wines at the end of AF and/or MLF. Information on grape variety, enological area, harvest year, and on other aspects of the winemaking processes, including yeast species dominating AF, is reported (Table 1). Samples were withdrawn daily from tanks soon after pumping over, refrigerated, and immediately transferred to laboratories for microbiological and chemical analyses.
Experimental fermentations.
Laboratory microvinifications were carried out in triplicate using grape juice with its skins, obtained by manual stemming and crushing of Sangiovese grape clusters harvested in 2009 and stored at −20°C until their use. All grape clusters were thawed on the same day, in order to avoid any chemical difference among grape juices, possibly caused by different time periods of storage at −20°C.
In order to arrange three distinct cases of dominant yeast populations, microvinifications were induced by inoculating the musts with specific yeast cultures: (1) Saccharomyces cerevisiae, (2) Kloeckera apiculata followed by S. cerevisiae after 3 days, and (3) Candida zemplinina followed by S. cerevisiae after 5 days. The yeast strains, one for each above-mentioned species, were previously isolated from wine fermentations and are included in the yeast culture collection of the Department of Agricultural Biotechnology, University of Florence, Italy. In all cases, yeast inocula originated from 1-day-old cultures, maintained at 30°C in yeast extract-peptone-dextrose agar (YEPD), to give an initial cell concentration of 106 cells/mL. This concentration, on the basis of a preliminary microbiological analysis of three representative grape clusters, was judged largely sufficient to guarantee the dominance of the inoculated yeasts. Fermentations were carried out in sterilized 250 mL Erlenmeyer flasks containing 230 g Sangiovese must and provided with Müller valves filled with concentrated sulfuric acid. The time courses of the fermentations were monitored by determining the weight loss caused by CO2 evolution. Once CO2 production ceased, the sugar content of the wines was determined by HPLC analysis and fermentation was considered complete when sugar concentration was less than 2 g/L. The experimental wines were stored at −20°C until chemical analysis.
Chemicals and analytical determinations.
All solvents were of HPLC quality and all chemicals of analytical grade (>99%); water was of MilliQ quality (Millipore, Billerica, MA). Yeast enumeration was obtained by surface spreading 0.1 mL samples of wine (diluted if necessary) onto plates of WL nutrient agar and lysine agar (Oxoid Limited, Hampshire, UK), supplemented with sodium propionate (2 g/L) and streptomycin (30 mg/L). The plates were incubated for 5 days at 30°C. Yeast isolates were identified by PCR-RFLP of rITS (Esteve-Zarzoso et al. 1999, Granchi et al. 1999, Sipiczki 2003).
Bacterial enumeration was performed by spreading dilutions of wine samples onto plates of MRS agar (Oxoid Limited) at pH 4.8, supplemented with pimaricin (50 mg/L; Sigma, St. Louis, MO) to suppress yeast growth and tomato juice broth (2 g/L; Difco Laboratories, Detroit, MI). The plates were incubated under microaerophilic conditions for 7 to 8 days at 30°C. The identification of the isolates was performed by both restriction analysis of the amplified 16S rDNA (PCR-ARDRA) and species-specific PCR according to published methods (Rodas et al. 2003, Zapparoli et al. 1998, respectively).
Free α-amino nitrogen content was determined by the NOPA procedure (Dukes and Butzke 1998) and ammonia by enzymatic test kit (Roche, Mannheim, Germany). Total acidity was determined according to the official method for wine analysis (EEC 2676/90).
HPLC separation and identification of individual anthocyanins were performed according to the Office International de la Vigne et du Vin method for analysis of anthocyanins in red wines (OIV 2009). The wine samples were injected in a ProStar 210 HPLC (Varian, Palo Alto, CA) equipped with a diode array detector and a reversed-phase column Polaris C18-A (5 μm particle, C18-A 250 × 4.6 mm; Varian), thermostated at 25°C. The anthocyanin profiles were determined by the relative proportions of the peak areas.
Glucose, fructose, and pyruvic acid concentrations in must and wine were determined by HPLC according to a published method (Granchi et al. 1998), using a MetaCarb H Plus Column (8 μm particle, 300 × 7.8 mm; Varian) and a ProStar 210 chromatograph equipped with a diode array detector at 210 nm and a refractive index detector, in series (Varian). All analytical data are the mean of two separate determinations.
Statistical analysis.
Experimental data were statistically treated by principal component analysis (PCA), Student’s unpaired t test, analysis of variance (ANOVA), and Tukey’s test. The statistical level of significance was set at p ≤ 0.01. All calculations were performed using Statistica software (version 7; StatSoft, Tulsa, OK).
Results and Discussion
The yeast populations dominating the AF of the investigated winemaking processes are reported (Table 1). When only S. cerevisiae (S.c.) is listed as dominant yeast species, that indicates that cell counts of the non-Saccharomyces populations remained well below 2 log units compared to S. cerevisiae and that their persistence in the transformation process was quite negligible. In contrast, when K. apiculata and/or C. zemplinina are reported together with S. cerevisiae, that indicates that growth of the two non-Saccharomyces yeast species occurred up to maximum populations of greater than 107 cfu/mL and that their persistence in the initial phases of the transformation process was quantitatively significant (at least until 30% degradation of initial sugar concentration). For the populations dominating MLF, at least in those wines that were analyzed to determine if the anthocyanin profile was affected by this secondary fermentation, bacterial populations other than Oenococcus oeni were not identified among the assayed isolates.
The data concerning the anthocyanin profiles of all the investigated monovarietal wines, analyzed after AF and/or after MLF, were preliminarily submitted to statistical analysis and results are reported as mean profiles of Sangiovese, Merlot, and Cabernet Sauvignon wines (Table 2). Cabernet Sauvignon and Merlot wines were used as references, as they have been extensively studied and their anthocyanin profiles are well-known in the literature (Burns et al. 2002, Revilla et al. 2001,de Villiers et al. 2004). Results showed that the mean anthocyanin profile of the Sangiovese wines was characterized by high percentages of malvidin-3-glucoside. The other anthocyanins ranged from 9 to 14%, without appreciable levels of acylated anthocyanins. This latter feature was in accordance with previous findings on young Sangiovese wines (Castellari et al. 1998, 2001). The mean anthocyanin profiles of Merlot and Cabernet Sauvignon wines showed a predominance of malvidin-3-glucoside, very low percentages of cyanidin-3-glucoside, and high relative abundances of acylated anthocyanins. This anthocyanin distribution is consistent with previous findings (Burns et al. 2002, Revilla et al. 2001,de Villiers et al. 2004). Compared to Merlot and Cabernet Sauvignon wines, the Sangiovese wines had a significantly higher percentage of cyanidin-3-glucoside, petunidin-3-glucoside, and peonidin-3-glucoside, but a much lower percentage of acylated anthocyanins.
A principal component analysis (PCA) of all the data was performed to determine whether the anthocyanin profiles of the wines grouped on any basis other than grape variety (enological area, vintage year, and/or winery). The first two principal components explained ~90% of the total variance (Figure 1A). The first principal component (factor 1) correlated positively with acetylated and coumaroylated anthocyanins and negatively with cyanidin-3-glucoside and peonidin-3-glucoside. The second principal component (factor 2) correlated negatively with malvidin-3-glucoside (Figure 1B). All Sangiovese wines clearly grouped on the left half of the plot and can readily be discriminated from Merlot and Cabernet Sauvignon wines on the basis of the relative abundance of the acylated anthocyanins. In contrast, at least for the wines here reported, PCA was unable to differentiate Cabernet Sauvignon from Merlot wines.
The Sangiovese wines did not group around the mean score. Instead, independently of either wine-producing area or harvest year of the grapes or winery, they were distributed along a longitudinal axis (Figure 1A), primarily due to a variable relative abundance of cyanidin-3-glucoside, peonidin-3-glucoside, and malvidin-3-glucoside. As a consequence of this wide variability, the anthocyanin profiles of the samples located at the extremities of this distribution were significantly different, as confirmed by comparing the three wine samples located at the two extremities and characterized by a high (lower extremity) and a low (upper extremity) malvidin-3-glucoside percentage, respectively. In particular, the anthocyanin profiles of these three Sangiovese wines were significantly different (p < 0.01) for delphinidin-3-glucoside, cyanidin-3-glucoside, and malvidin-3-glucoside (Table 3). This variability in the anthocyanin profile of the Sangiovese wines and the well-ordered distribution of the samples in the PCA plane suggested a more detailed analysis of the results. At first, in order to establish whether the PCA distribution of the Sangiovese wines was affected by the yeast ecology of fermentation, the data of the Sangiovese wines were subdivided into four groups, according to both the fermentation type (spontaneous or inoculated) and the yeast populations dominating the winemaking process (Table 1). Since the statistical analysis (Table 4) did not indicate significant difference in the malvidin-3-glucoside percentage of the groups, the yeast ecology of winemaking, at least when it is described at species level, does not appear to be responsible for the distribution of the wine samples observed in the PCA plane.
The vitisin A percentage of the Sangiovese wines was significantly higher in those wines where a spontaneous AF with numerically high populations of C. zemplinina occurred. To verify if this higher vitisin A percentage was really dependent on the growth and activity of C. zemplinina, a laboratory experiment was performed using Sangiovese grape musts inoculated with either a pure culture of S. cerevisiae or with sequential cultures of K. apiculata and S. cerevisiae or with sequential cultures of C. zemplinina and S. cerevisiae. The anthocyanin profiles of the resulting experimental wines presented, as the only significant difference, higher percentages of vitisin A in the wines made with sequential cultures of C. zemplinina and S. cerevisiae (Table 5), suggesting a role of C. zemplinina in vitisin A formation. Indeed, HPLC analysis of the experimental wines demonstrated significantly higher pyruvic acid concentration in the wines fermented with sequential cultures of C. zemplinina and S. cerevisiae (110.3 ± 10.9 mg/L, expressed as mean ± SD) than in the wines fermented with a pure culture of S. cerevisiae (12.9 ± 0.52 mg/L) or the wines fermented with sequential cultures of K. apiculata and S. cerevisiae (21.3 ± 2.55 mg/L). Hence, the higher vitisin A percentage of the wines made with sequential cultures of C. zemplinina and S. cerevisiae seems to be related to a higher pyruvic acid-producing capability of this microbial combination.
The influence of MLF on the anthocyanin profile of the Sangiovese wines was investigated by considering the analytical data of the 11 commercial vinifications analyzed both before the onset and after completion of spontaneous MLF. Results did not indicate significant differences between the anthocyanin profile characterizing a wine before the onset of MLF and the profile of the same wine after MLF completion (Table 6), showing that the anthocyanin profile of a red wine is not significantly modified by O. oeni populations while performing MLF.
Conclusion
All the Sangiovese wines considered in this study—produced at commercial scale in different enological areas of Tuscany with a variable microbial ecology (inoculated or spontaneous AF, absence or presence of significant populations of non-Saccharomyces yeasts, before onset or after completion of MLF by LAB)—showed an anthocyanin profile characterized by the prevalence of malvidin-3-glucoside and the trace presence or even absence of acylated anthocyanins. This anthocyanin pattern, also shared by all the wines produced in the laboratory by experimental vinifications, reproduces the peculiar profile of Sangiovese grapes, which have few acylated anthocyanins. Consequently, the profile could be used to identify distinctive features of Sangiovese monovarietal wines or, at least, to distinguish Sangiovese wines from wines made with other grape varieties, such as Merlot or Cabernet Sauvignon, that are known to possess significant proportions of acylated anthocyanins.
However, despite this distinctive feature, the anthocyanin profiles of the Sangiovese wines showed a wide variability, mostly dependent on the relative abundances of cyanidin-3-glucoside, peonidin-3-glucoside, and malvidin-3-glucoside. This variability, independent of wine-producing area, harvest year of the grapes, or winery, was also unaffected by the microbial ecology of the vinification process, at least if microbial ecology is described at species level. However, the occurrence of microbial strains possessing specific adsorbent properties, capable of modifying anthocyanin profile according to the observed distribution of the samples in the PCA plane, cannot be ruled out.
Moreover, the observed variability in the anthocyanin profile of the Sangiovese wines could also depend on factors other than those taken into consideration in this study, such as vineyard management and maximum temperature reached during AF. Indeed, it is widely known that anthocyanin profiles of shaded grapes are characterized by high percentages of dioxygenated anthocyanins, cyanidin-3-glucoside and peonidin-3-glucoside, as found in those wine samples located at the upper extremity of the PCA plot (i.e., samples with low malvidin-3-glucoside percentages). On the other hand, in that malvidin-3-glucoside is more resistant to oxidation than the other anthocyanins, a lower percentage of both delphinidin-3-glucoside and cyanidin-3-glucoside, associated with a high relative abundance of malvidin-3-glucoside as found in those wine samples located at the lower extremity of the PCA plot (i.e., samples with high malvidin percentages), could be the consequence of high temperatures, known to accelerate anthocyanin oxidative processes and possibly reached during AF in winemaking processes without temperature control.
Acknowledgments
Acknowledgments: Research funded by Fondazione Ente Cassa di Risparmio di Firenze.
- Received May 1, 2011.
- Revision received July 1, 2011.
- Accepted August 1, 2011.
- Published online December 1969
- © 2011 by the American Society for Enology and Viticulture