Quality and Composition of Airén Wines Fermented by Sequential Inoculation of Lachancea thermotolerans and Saccharomyces cerevisiae

Traditionally, Saccharomyces cerevisiae is the yeast used widely for winemaking. However, grapes are not sterile media and there are many other yeast species with plenty of potential to solve new oenology challenges that must be studied. Several research groups have studied non-Saccharomyces yeast applications (1,2) in diff erent grape varieties such as Sauvignon blanc (3,4), Chenin blanc (4), Chardonnay (4–6), Amarone (7), Muscat (8), Muscat d’Alexandrie (9), Debina (10), Macabeo (11,12), Folle blanche (13), Bobal (14), Alvarinho, Loureiro, Trajadura, Pedernã, Azal Branco, Avesso (15), Airén (16,17), Pedro Ximenez (18), Sangiovese (19), Pinot noir (20), Emir (21,22), Syrah (23–26), Tempranillo (27,28) and Riesling (29). In most cases improvements in wine quality were reported. The presence of wild non-Saccharomyces yeasts in fermentations was traditionally associated with high levels of acetic acid and other off -fl avours. Nevertheless, nowadays researchers and winemakers are aware of the positive infl uence of non-Saccharomyces yeasts on wine aroma complexity (1,2,29–40). The development of multistarter fermentation with Saccharomyces cerevisiae as a binding partner has been proposed to overcome the shortcomings of alcoholic fermentation with non-Saccharomyces yeasts. Mixed fermentations are of interest because of some enzymatic properties (glycosidases, β-lyase, etc.), ethanol reduction and the release of some interesting metabolites such as glycerol, pyruvic acid and mannoproteins, among others (41–44).


Introduction
In recent years, global climate change has set a trend towards an increase in sugar content and a decrease in the acidity of grape juices.Microbiological acidifi cation can play an essential role in satisfying the growing wine market demand for quality wines.
This study aims to enhance Airén wine quality.This Spanish variety is considered as a neutral and very productive grape but is usually associated with low-quality wine.However, this variety is the most planted in Spain.Most Airén vineyards are located in the centre of the southern Spain.This area is considered to be a warm semi-desert region, where high sugar contents and lack of acidity in wine are the main problems.Therefore, Lachancea thermotolerans 617 was selected among other non-Saccharomyces yeasts in this study to perform combined fermentations with S. cerevisiae in order to increase the acidity and quality of Spanish Airén wine.

Microorganisms
The following yeast strains were used for the experimental fermentation of the studied Airén must: Saccharomyces cerevisiae 87 (CECT 12512; Spanish Type Culture Collection, Valencia, Spain) and Lachancea thermotolerans 617 (CECT 12672; Spanish Type Culture Collection).

Vinifi cation
Grapes of Airén cultivar (Vitis vinifera L.), grown in El Socorro experimental vineyard (Madrid, Spain) were used in the fermentations.Using a microvinifi cation method similar to that described in scientifi c literature (60), 3.9 L of sterilised must (115 °C, 15 min) were placed in 4.9-litre glass fermentation vessels, leaving enough space for the emission of carbon dioxide.No sulphur dioxide was added to any vessel.Sugar concentration was 244.51 g/L, pH=3.68,primary amino nitrogen (PAN; Biosystems S.A., Barcelona, Spain) 177 mg/L, and lactic and acetic acids below 0.1 g/L.A concentration of 0.4 g/L of Actimax Natura (Agrovin S.A., Alcázar de San Juan, Spain), inactivated autolyzed yeasts naturally rich in amino acids, was added to provide nutrition for the media.Three assays were performed (all in triplicate): (i) inoculation of the must with S. cerevisiae 87 alone (SC; 100 mL containing 1.18•10 7 CFU/mL), (ii) inoculation with S. cerevisiae 87 (1.18•10 7 CFU/mL) and L. thermotolerans 617 (100 mL containing 2.95•10 7 CFU/mL) together (mixed fermentation: LT×SC), and (iii) inoculation with L. thermotolerans 617 (100 mL containing 2.27•10 7 CFU/mL) followed by S. cerevisiae 87 (100 mL containing 10 7 CFU/mL) 96 h later (sequential fermentation: LT...SC).Yeast inocula were produced using 100 mL of sterilised must with 1 mL of yeast extract peptone dextrose (YEPD; Pronadisa, Madrid, Spain) liquid medium (61), in the concentration of 10 6 CFU/mL (determined using a counting chamber).To reach this population, 100 μL of each yeast suspension were cultivated in 10 mL of YEPD at 25 °C for 24 h.This procedure was repeated successively three times before the fi nal inoculation of 1 mL of the suspension.All inocula were prepared in 250-mL fl asks fi lled with 98 % H 2 SO 4 (Panreac, Barcelona, Spain), which allowed the release of CO 2 while avoiding microbial contamination (62), and sealed with a 14-cm Muller valve (Alamo, Madrid, Spain).The temperature was maintained at 25 °C for 48 h.The development of inocula proceeded without aeration, oxygen injection or agitation.All fermentation processes, which were done in triplicate, were carried out at 25 °C.When the sugar concentration fell below 3 g/L, the wines were racked and stabilised for 7 days at 4 °C.The wine was then bott led, and a concentration of 40 mg/L of sulphur dioxide in the form of potassium disulfi te was added.Sealed bott les were placed horizontally in a climate chamber at 4 °C for three weeks until the sensory evaluation.

Analytical determinations of non-volatile compounds
Glucose and fructose, l-lactic acid, acetic acid, glycerol, pyruvic acid, acetaldehyde, l-malic acid and primary amino nitrogen were all determined using a Y15 enzymatic autoanalyzer (Biosystems S.A.) with corresponding kits.Ethanol, pH, free SO 2 and total SO 2 profi le were determined following the methods described in the Compendium of International Methods of Analysis of Wines and Musts (63).

Growth kinetics during microvinifi cation
During fermentations, aliquots were taken periodically under aseptic conditions and further tenfold dilutions were made serially.Growth kinetics was monitored by plating 100 μL of the appropriate dilution on lysine medium (Oxoid, Basingstoke, UK) for counting non-Saccharomyces yeasts (64) and YEPD medium (Pronadisa) for total yeast counts (61).Colonies were counted aft er growth at 30 °C for 48-72 h.

Analytical determination of volatile compounds
The concentration of volatile compounds, all of which infl uence wine quality, was measured at the end of alcoholic fermentation by gas chromatography using an Agilent Technologies 6850 gas chromatograph with a fl ame ionisation detector (Hewlett Packard, Palo Alto, CA, USA) (65), calibrated with 4-methyl-2-pentanol (Fluka, Sigma--Aldrich Corp., Buchs, Switzerland) as an internal standard.Gas chromatography standards (Fluka, Sigma-Aldrich Corp.) were used to provide standard patt erns.Higher alcohols were separated according to the Compendium of International Methods of Analysis of Wines and Musts (63), with the detection limit of 0.1 mg/L.Minor compounds were quantifi ed using gas chromatography-mass spectrometry as described by Lopez et al. (66) with the modifi cations introduced by Loscos et al. (67).

Analytical determination of amino acids
The amino acids were analysed using a Jasco (Tokyo, Japan) ultra-high-performance liquid chromatograph (UH-PLC) series X-LCTM, equipped with a fl uorescence detector 3120-FP.Gradients of solvent A (methanol/acetonitrile 50:50, by volume) and B (sodium acetate/tetrahydrofuran 99:1, by volume) were used in a C18 (HALO ® , Wilming-ton, DE, USA) column (100 mm×2.1 mm; particle size 2.7 μm) as follows: 90 % B at 0.25 mL/min, from 0 to 6 min; 90-78 % linear gradient B at 0.2 mL/min, from 6 to 7.5 min; 78 % B from 7.5 to 8 min, 78-74 % linear gradient B at 0.2 mL/min, from 8 to 8.5 min, 74 % B at 0.2 mL/min, from 8.5 to 11 min, 74-50 % linear gradient B at 0.2 mL/min, from 11 to 15 min, 50 % B at 0.2 mL/min, from 15 to 17 min, 50-20 % linear gradient B at 0.2 mL/min, from 17 to 21 min, 20-90 % linear gradient B at 0.2 mL/min, from 21 to 25 min and re-equilibration of the column from 25 to 26 min to the initial gradient conditions.The scanning range for the detection of amino acids was 340-455 nm.Amino acids were quantifi ed by comparison against their external standards, and diff erent acids were identifi ed by their retention times.

Analytical determination of biogenic amines
The biogenic amines were analysed using a Jasco UHPLC chromatograph series X-LCTM, equipped with a fl uorescence detector 3120-FP.Gradients of solvent A (methanol/acetonitrile, 50:50, by volume) and B (sodium acetate/tetrahydrofuran, 99:1, by volume) were used in a C18 (HALO ® ) column (100 mm×2.1 mm; particle size 2.7 μm) as follows: 60 % B at 0.25 mL/min, from 0 to 5 min; 60-50 % linear gradient B at 0.25 mL/min, from 5 to 8 min; 50 % B from 8 to 9 min, 50-20 % linear gradient B at 0.2 mL/min, from 9 to 12 min, 20 % B at 0.2 mL/min, from 12 to 13 min, 20-60 % linear gradient B at 0.2 mL/min, from 13 to 14.5 min, and re-equilibration of the column from 14.5 to 17 min to the initial gradient conditions.The scanning range for the detection of biogenic amines was 340-420 nm.Biogenic amines were quantifi ed by comparison against their external standards, and diff erent amines were identifi ed by their retention times.

Sensory analysis
The obtained wines were assessed using a blind test by a panel of 15 experienced wine tasters, all staff members of the Chemistry and Food Technology Department of the Polytechnic University of Madrid (Madrid, Spain) and the Accredited Oenology Laboratory of Haro (Haro, Spain).Following consistent terminology by consensus, fi ve aromas and fi ve taste att ributes were chosen to describe the wines.The panellists used an unstructured scale, with scores ranging from 0 (no character) to 6 (very strong character), to rate the intensity of the 11 att ributes.

Statistical analysis
PC Statgraphics v. 5 soft ware (Graphics Soft ware Systems, Rockville, MD, USA) was used for statistical analyses.The signifi cance was set to p<0.05 for the ANOVA matrix F value.The mean values were compared using multiple range test.the number of L. thermotolerans 617 cells started to decline fast.Other authors reported previously fermentation kinetics of other non-Saccharomyces strains, in which the presence of non-Saccharomyces strains was also observed during the early stages of fermentation.In this trial L. thermotolerans 617 strain disappeared on day 8 in the sequential (LT...SC) fermentation (Fig. 1).This can be explained by the higher fermentation activity of this species compared to other low-fermenting non-Saccharomyces strains.Some S. cerevisiae strains were also reported to secrete antimicrobial peptides that inhibit non-Saccharomyes yeast growth (68).This could explain the early disappearance of L. thermotolerans once S. cerevisiae was inoculated, even though it has been reported to tolerate up to 9 % (by volume) of ethanol when it ferments on its own (55).In this trial, the LT...SC fermentation was the best option.In the case of the LT×SC fermentation, L. thermotolerans disappeared fast so acidifi cation was not completed.Cell fl occulation or loss of viability can explain the observed reduction in cell numbers during fermentation.

Sugar consumption kinetics and alcohol production
The Saccharomyces cerevisiae 87 fermenting on its own (SC) and in the LT×SC fermentation consumed the sugar the fastest (Fig. 2).Fermentation time varied from 10 to 14 days and fi nal alcohol content varied from 13.91 to 14.36 % (by volume).The ethanol content was lower in the LT... SC fermentation (Table 1).The sugar consumption results analysed in this work (Fig. 2) are in agreement with the lower fermentation activity of Lachancea spp.compared with S. cerevi siae (55), due to the fact that in the last stages of fermen tation only S. cerevisiae was detected.Several authors question the usefulness of non-Saccharomyces yeast in the production of lower volume fractions of alcohol in wines (43,69).These previous results are in agreement with the lower fi nal alcohol content of the wines produced in the sequential fermentations involving Lachancea thermotolerans 617 (Table 1).However, in our case the alcohol reduction was about 0.4 % (Table 1).

Acetic acid metabolism
Fig. 3 shows the kinetics of acetic acid release.Acetic acid concentration varied from 0.31 to 0.39 g/L (Table 1).LT...SC fermentation produced the lowest fi nal acetic acid concentration.SC and LT×SC fermentations had similar fi nal acetic acid content of about 0.38 g/L (Fig. 3).One of the problems raised by winemakers is the excessive increase of acetic acid in wines with high presence of non--Saccharomyces yeasts (1).However, previous experiments with L. thermotolerans reported signifi cant reduction in final volatile acidity in sequential fermentations of 0.25 (19), 0.06 (56), 0.2 (42) and 0.08 g/L (70).Our results confi rm an additional decrease in this compound related to the presence of L. thermotolerans (Fig. 3; Table 1).Never-  theless, acetic acid concentration in all fermentations was not excessive and it did not aff ect wine quality negatively.
The results show that a controlled use of L. thermotolerans in sequential fermentations can cause a decrease of acetic acid production.

l-lactic acid metabolism
Fig. 4 reports that only the fermentations involving Lachancea thermotolerans 617 produced l-lactic acid.The results varied from 0.24 g/L in LT×SC to 3.18 g/L in LT...SC (Table 1).Other authors obtained signifi cant acidifi cations using combined microbiological cultures of L. thermotolerans and S. cerevisiae with the main objective of acidifying musts that were low in titratable acidity.Previously obtained values such as 3.42 g/L (19) were similar to the ones reported in this work.In other cases, acidifi cation was higher; up to 5.13 g/L (56) has been reported depending on diff erent trial conditions.The production of l-lactic acid is linked to the viable cell content (70).LT...SC fermentation proved to be the best option for acidifying wine in this study (Fig. 4; Table 1).In the case of LT×SC fermentation, the acidifi cation was signifi cantly lower due to the fast Saccharomyces growth, which impeded a higher acidifi cation by L. thermotolerans.

l-malic acid metabolism
Only the fi nal malic acid content in the SC fermentation was lower than in the other fermentations (Table 1); the maximum malic acid reduction rates of 17.65 % in SC, 14.29 % in LT×SC and 9.25 % in LT...SC fermentation from the initial concentration of 1.19 g/L were detected.The slight decrease in malic acid content observed in the fermentations (Table 1) is in agreement with other authors who confi rmed that malic acid can be metabolised by several yeast species (44,52,57) at levels lower than 20 %, unless Schizosaccharomyces genus is involved.

Glycerol production
The glycerol content in LT...SC fermentation was higher than those observed in SC and LT×SC fermentations (Table 1).Final levels of glycerol varied from 7.11 to 7.55 g/L (Table 1).Increased glycerol content is described as one of the main contributions of non-Saccharomyces strains to wine quality (71) because it contributes positively to the mouthfeel.L. thermotolerans has been described before in literature as a higher glycerol producer than S. cerevisiae, reporting increases of about 0.69 (19) and 0.93 g/L (56).However, some authors have reported that an increase in glycerol production is usually related to an increase in acetic acid production (72), which can be detrimental to wine quality.Our results confi rm that this fact seems to be irrelevant in the case of LT…SC fermentation.

Pyruvic acid production
Maximum pyruvic acid production was observed between the second and fourth day, reaching 128 and 149 mg/L, respectively (Fig. 5) during the fermentation of Saccharomyces cerevisiae 87 alone (SC) or LT×SC fermentation.LT...SC fermentation had higher values with a maximum concentration of pyruvic acid of 172.36 mg/L on day 6.Previous studies on the production of pyruvic acid by S. cerevisiae strains reported maximum values of 60-132 mg/L aft er 4 days of fermentation (52).Similar values were obtained in the present study in SC and slightly higher in LT×SC fermentation (Fig. 5).Nevertheless, the LT...SC fermentation obtained signifi cantly higher levels, but not as high as those described for the genus Schizosaccharomyces (52).The concentrations of pyruvic acid and glycerol could indicate that L. thermotolerans possesses a highly active glyceropyruvic pathway (73).

Acetaldehyde production
The fermentations involving L. thermotolerans 617 produced less acetaldehyde, with values that varied from 27 in LT…SC to 35 mg/L in LTx×SC (Table 1).SC fermen- tation produced more acetaldehyde than the others, with a fi nal concentration of 39.00 mg/L (Table 1).Acetaldehyde is produced from the yeast metabolism of sugars and it is partly re-utilized (74).Although SC fermentation produced more acetaldehyde than the others (Table 1), all fi nal values were under the sensory threshold of 100-125 mg/L (75).

Volatile aroma
Isoamyl alcohol, ethyl octanoate and isoamyl acetate were formed in higher total concentrations during SC and LT×SC fermentations (Table 2).On the other hand, fi nal concentrations of ethyl lactate, 2-phenylethanol and 2-phe nylethyl acetate (Table 2) up to 5.98, 3.92 and 0.16 mg/L higher, respectively, were reported in LT…SC than in SC fermentation.Other authors have described non--Saccharomyces yeasts as weaker producers of higher alcohols than Saccharomyces cerevisiae (10,11,19,46,76).LT…DC fermentation produced the most 2-phenylethanol (Table 2).Other authors have reported higher production of 2-phenyl ethanol and ethyl lactate by L. thermotolerans than by S. cerevisiae (19), by up to 7.92 and 14.34 mg/L, respectively.L. thermotolerans has also been reported as a weaker ethyl acetate producer than S. cerevisiae (19).

Amino acids and biogenic amines
Higher fi nal levels of histidine, glycine and leucine were obtained in SC and LT×SC fermentations than in LT…SC fermentation (Table 3).LT…SC fermentation had higher fi nal levels of alanine, lysine and serine (Table 3).The fi nal concentrations of each biogenic amine were always lower than 1 mg/L (Table 4).Diff erences in the amino acid patt erns among the diff erent fermentations were found, but they could not be related to the aroma of the Airén wines.Diff erent autolysis behaviour might be the reason for this.A histamine value of 2 mg/L is considered a limiting factor (77) in some countries due to food safety legislation.Our results prove that L. thermotolerans does not produce higher levels of biogenic amines than S. cerevisiae.However, most biogenic amines are produced during malolactic fermentation and wine ageing (78).Nevertheless, the lower concentration of histidine (precursor of histamine) found during LT…SC fermentation (Table 3) contributes to reducing the potential risk of histamine formation by bacterial metabolism.Even though no significant diff erences were found in fi nal biogenic amine contents, other authors have reported reductions of histamine

Sensory evaluation
Wines produced in LT…SC fermentation trials had bett er sensorial properties and general acidity (Fig. 6).However, SC and LT...SC fermentations scored highest in sweetness (Fig. 6).This can be easily explained by the elevated l-lactic acid production by L. thermotolerans.Lack of acidity is a common fault described for Spanish Airén grape variety when compared to other European varieties.Although the wines obtained in SC and LT×SC fermentations were evaluated as sweeter than those in the LT...SC fermentation, all fi nal wines were considered dry from a chemical point of view (Table 1).This perception could be explained due to the diff erent balance between the acidity and sweetness.

Conclusions
The comparison of the results between the fermentation trials showed diff erences in several analysed parameters and the positive infl uence of the studied Lachancea thermotolerans yeast strain on Airén wine quality.Finally, sequential fermentation with L. thermotolerans and Saccharomyces cerevisiae remains the best option, as it considerably increased acidity and complexity of the studied neutral grape variety.

Fig. 1
Fig. 1 shows the development of diff erent yeast strains during fermentation.Fermentation time varied

Table 1 .
Analytical results for the wines produced by diff erent fermentations

Table 3 .
Concentrations of amino acids determined aft er diff erent fermentationsResults represent the mean value±S.D. of three replicates.Mean value in the same row with the same lett er are not signifi cantly diff erent (p<0.05)SC=fermentation with Saccharomyces cerevisiae 87 alone, LT×SC= mixed fermentation with Lachancea thermotolerans 617 and S. cerevisiae 87, LT…SC=sequential fermentation with L. thermotolerans 617 followed by S. cerevisiae 87

Table 4 .
(6)genic amine concentration in the studied fermentationsResults represent the mean value±S.D. of three replicates.Mean values in the same row with the same lett er are not signifi cantly diff erent (p<0.05)SC=fermentation with Saccharomyces cerevisiae 87 alone, LT×SC= mixed fermentation with Lachancea thermotolerans 617 and S. cerevisiae 87, LT…SC=sequential fermentation with L. thermotolerans 617 followed by S. cerevisiae 87 of up to 2.2 mg/L during alcoholic fermen tation with the non-Saccharomyces species Hanseniaspora vineae(6).