Journal of Microbiological Methods 119 (2015) 176–179

Contents lists available at ScienceDirect

Journal of Microbiological Methods

journal homepage: www.elsevier.com/locate/jmicmeth

A new method for monitoring the extracellular proteolytic activity of

wine yeasts during alcoholic fermentation of grape must
Laura Chasseriaud a,b, Cécile Miot-Sertier a,c, Joana Coulon b, Nerea Iturmendi b, Virginie Moine b,
Warren Albertin a, Marina Bely a,⁎
a
Univ. de Bordeaux, ISVV, EA 4577, Unité de Recherche Œnologie, 33140 Villenave d'Ornon cedex, France
b
BioLaffort, 126 Quai de la Souys, 33100 Bordeaux, France
c
INRA, ISVV, USC 1219 Œnologie, 33140 Villenave d'Ornon cedex, France

a r t i c l e i n f o a b s t r a c t

Article history: The existing methods for testing proteolytic activity are time consuming, quite difficult to perform, and do not
Received 19 October 2015 allow real-time monitoring. Proteases have attracted considerable interest in winemaking and some yeast spe-
Received in revised form 30 October 2015 cies naturally present in grape must, such as Metschnikowia pulcherrima, are capable of expressing this activity.
Accepted 30 October 2015
In this study, a new test is proposed for measuring proteolytic activity directly in fermenting grape must, using
Available online 31 October 2015
azocasein, a chromogenic substrate. Several yeast strains were tested and differences in proteolytic activity
Keywords:
were observed. Moreover, analysis of grape must proteins in wines revealed that protease secreted by
Proteolytic activity Metschnikowia strains may be active against wine proteins.
Azocasein © 2015 Published by Elsevier B.V.
Wine
Alcoholic fermentation
Metschnikowia spp.

1. Introduction
Current testing methods for proteolytic activity use solid or liquid
laboratory media. Milk, casein, or gelatine plates are frequently used
and proteolytic activity is estimated by the appearance of a clearance
zone around the colony (Charoenchai et al., 1997; Fernández et al.,
2000; Mateo et al., 2015; Reid et al., 2012; Strauss et al., 2001). This re-
quires quite a long incubation time and, moreover, the results obtained
by these methods are qualitative and do not differentiate among the
various strains' proteolytic activity. Quantifying proteolytic activity
requires the use of laboratory buffers, such as phosphate or citrate–
phosphate, containing proteins, generally BSA (Lagace and Bisson,
1990; Mateo et al., 2015). The presence of active proteases leads to an
increase in the optical density of the solution. The Cd-ninhydrin method
(Maturano et al., 2012; Mendoza et al., 2007) is also used to measure
proteolytic activity. Cells are incubated with proteins and Cd-
ninhydrin reagent in the presence of citrate-citric acid buffer. Proteolyt-
ic activity can be determined by measuring optical density. Oxidized in-
sulin is also used as a protease substrate for measuring proteolytic
activity in wine. NH2 products released by protein hydrolysis are mea-
sured by ninhydrin (Humbert-Goffard, 2003). In these methods, pH is
not always adjusted to a value representative of white wine (between
⁎ Corresponding author at: Univ. de Bordeaux, ISVV, EA 4577, Unité de Recherche
Œnologie, 210 chemin de Leysotte, 33140 Villenave d'Ornon cedex, France.
E-mail address: laura.chasseriaud@u-bordeaux.fr (M. Bely).
3.0 and 3.5). Moreover, while incubation times are shorter than those
required by plate methods, this analysis remains time-consuming and
quite difficult to perform.
Wine haze forms when proteins become unstable and insoluble. The
mechanism is not entirely understood, but is hypothesized to result
from thermal denaturation of proteins under unfavourable storage
conditions. It may be induced by several physicochemical factors, in-
cluding pH, ethanol concentration, etc. (Dawes et al., 1994; Sarmento
et al., 2000). This phenomenon reduces the commercial value of the
wine. Some pathogen-related proteins (β-glucanases, chitinases, and
thaumatin-like proteins) are responsible for haze formation. They are
characterised by a molecular mass between 13 and 30 kDa and an iso-
electric point between 4.1 and 5.8 (Hsu and Heatherbell, 1987;
Lamikanra and Inyang, 1988; Waters et al., 1992). Bentonite is frequent-
ly used to reduce the risk of haze formation in wine, but this treatment
presents some disadvantages: loss of volume, non-optimum efficiency
against all protein classes at a single concentration, the use of mineral
material non-intrinsic to wine, and the elimination of some aroma com-
pounds. Wine treatments with acid yeast proteases may offer a microbi-
al alternative, making it possible to reduce the use of bentonite (Jolly
et al., 2003; Lagace and Bisson, 1990; Pocock et al., 2003). Besides reduc-
ing haze formation, proteases release peptides and amino acids from
proteins, thus increasing the assimilable nitrogen for yeast in must,
with a positive impact on wine aroma. Proteases may also be useful
for releasing mannoproteins in wine via lees autolysis, thus enhancing
wine protein stability (Moine-Ledoux and Dubourdieu, 1999). Proteases

http://dx.doi.org/10.1016/j.mimet.2015.10.025
0167-7012/© 2015 Published by Elsevier B.V.
 L. Chasseriaud et al. / Journal of Microbiological Methods 119 (2015) 176–179 177

need to be active at the low pH (3.0–3.5) of wine, in the presence of SO2, fermentation in h), maximum fermentation rate (g·L−1·h−1), and
and at winemaking temperatures. maximum CO2 release (g·L−1).
Several yeast genera, including Saccharomyces, Hanseniaspora,
Candida, Metschnikowia, Pichia, Torulaspora, and Issatchenkia are 2.3. Cell enumeration
frequently isolated from fermenting grape must (Baleiras Couto et al.,
2005; Combina et al., 2005; Fleet et al., 2002; Li et al., 2010; Zott et al., Yeast growth during fermentation was monitored by plate counting
2008). As the environment changes (increase in ethanol content, on a YPD-based medium: 1% yeast extract (w/v), 1% Bacto peptone
presence of inhibitors, competition for nutrients, etc.), Saccharomyces (w/v), 2% glucose (w/v), and 2% agar (w/v). Plates were incubated
cerevisiae species become predominant and complete the alcoholic at 24 °C.
fermentation.
S. cerevisiae is not known for abundant production of extracellular
2.4. Proteolytic activity
proteases. Recently, Younes et al. (2013) characterised some prote-
ases but they are not active under oenological conditions. Several
A chromogenic protease substrate, azocasein (Sigma Aldrich), was
studies have shown that some non-Saccharomyces species, particu-
used to monitor proteolytic activity directly in fermenting grape must.
larly M. pulcherrima, have an interesting potential in this regard
Azocasein proteolysis released a free dye into the supernatant, which
(Charoenchai et al., 1997; Fleet, 1992; Theron and Divol, 2014).
was quantified by measuring optical density at 440 nm. Several concen-
This article proposes a simple method for assessing and monitoring
trations of azocasein stock solution, prepared in NaOH 0.1 M, were test-
the proteolytic activity of yeasts during the alcoholic fermentation of
ed, as well as various final concentrations in bioreactors, to determine
grape must.
the optimum quantity of azocasein needed to be in excess of the
substrate without markedly increasing must pH (data not shown).
2. Materials and methods
The optimum concentration for stock solution was determined to
be 20 mg·L− 1, giving a final concentration of 1.5 mg·L− 1 in the
2.1. Yeast strains
bioreactors.
Samples were taken throughout alcoholic fermentation. The prote-
Seven yeast strains were used: 4 M. pulcherrima, 1 Metschnikowia spp.
olysis reaction was stopped with trichloroacetic acid (10% final concen-
(99% identity with Metschnikowia andauensis), 1 S. cerevisiae, and 1
tration, v/v, Sigma). Samples were centrifuged at 21, 693 g for 10 min
Torulaspora delbrueckii (Table 1).
and then the optical density of the supernatant was measured to evalu-
ate the proteolytic activity of the species tested. Initial assays were
2.2. Preculture and alcoholic fermentation conditions
carried out with or without azocasein to check that this substrate did
not impact yeast growth or alcoholic fermentation.
2.2.1. Precultures
Except for S. cerevisiae and T. delbrueckii, two preculture steps were
2.5. Analysis of grape must proteins
necessary before alcoholic fermentation. First, strains were grown in
YPD-based medium containing 1% yeast extract (w/v), Difco Laboratories,
At the end of alcoholic fermentation, 50 mL samples from S. cerevisiae,
Detroit, MI), 1% Bacto peptone (w/v, Difco), and 2% glucose (w/v) at 24 °C
Metschnikowia spp. CRBO L0563, M. pulcherrima IWBT Y1123, and
for 24 h. Then, the cultures were transferred to half-diluted grape must
M. pulcherrima Y6259 were centrifuged at 6000 rpm for 5 min. The mo-
with agitation at 24 °C for 24 h. Culture in half diluted grape must was
lecular weights of the supernatant macromolecules were determined by
only necessary for S. cerevisiae and T. delbrueckii.
HPLC on a TSKgel® G2000 SW column (Phenomenex), using the proto-
col described by Dubourdieu et al. (1986) for the fractionation of mole-
2.2.2. Alcoholic fermentations
cules from 10 to 70 kDa. Three molecular masses were separated by
Alcoholic fermentations were carried out in 125 mL of Sauvignon
retention time, corresponding to three fractions: P1 (N50 kDa), P2
Blanc grape must from the Bordeaux region (reducing sugars: 203 g/L,
(40 kDa) and P3 (b30 kDa). The grape must proteins responsible for
pH: 3,3) at 18 °C (usual temperature for white grape must fermenta-
haze formation are characterised by molecular masses between 13 and
tion), in 140 mL flasks, with agitation. All fermentations were performed
30 kDa (Waters et al., 1992), associated with the P3 fraction.
in triplicate. The different strains were inoculated as pure cultures at
5 × 106 viable cells/mL. Fermentation kinetics were monitored by regular
measurements of the weight loss due to CO2 release. Several growth and 2.6. Data analysis
fermentation parameters were calculated for each species: growth rate
(division·h−1), maximum population size (CFU·mL−1), fermentation R software was used for statistical analysis. Variance heteroscedasticity
lag phase (time between yeast inoculation and the beginning of alcoholic made it possible to apply the Kruskal–Wallis test to determine the differ-
ence between the growth and alcoholic fermentation parameters with
and without azocasein and the effect of M. pulcherrima protease on
Table 1
Strains used in alcoholic fermentation. grape must protein.

Species Strain Collection

3. Results
Metschnikowia CRBO L1329 Centre de Ressources Biologiques
pulcherrima Oenologique, Bordeaux, France
Y6259 Northern Regional Research Laboratory
3.1. Azocasein did not impact yeast growth or alcoholic fermentation
NZ366 Centre de recherche Pernod-Ricard,
Créteil, France In this part we focused on the effect of azocasein on the growth
IWBT Y1123 Institute for Wine Biotechnology, and alcoholic fermentation kinetics of several wine yeast species:
Stellenbosch University, South Africa
S. cerevisiae (X5), T. delbrueckii (Alpha), M. pulcherrima (IWBT 1123),
Metschnikowia spp. CRBO L0563 Centre de Ressources Biologiques
Oenologique, Bordeaux, France and Metschnikowia spp. (CRBO L0563). Several growth and alcoholic
Saccharomyces Zymaflore X5 Laffort, Bordeaux, France fermentation parameters were calculated for each species, as described
cerevisiae in materials and methods (Table 2). All fermentations with S. cerevisiae
Torulaspora Zymaflore Alpha were completed whereas, as expected, T. delbrueckii and Metschnikowia
delbrueckii
cultures stopped at 85% and 65%, respectively.
178 L. Chasseriaud et al. / Journal of Microbiological Methods 119 (2015) 176–179

Table 2
Growth and alcoholic fermentation parameters with and without azocasein.

Growth rate Maximum population size Fermentation lag Maximum fermentation rate Maximum CO2 released
(division·h−1) (cell·mL−1) phase (h) (g·L−1·h-1) (g·L−1)

S. cerevisiae X5 Without azocasein 0.14 ± 0.01 1.1 × 108 ± 1.2 × 106 14.0 ± 0.2 1.07 ± 0.02 91.6 ± 2.76
With azocasein 0.10 ± 0.04 1.1 × 108 ± 1.4 × 106 11.2 ± 0.1 1.02 ± 0.01 91.7 ± 0.03
T. delbrueckii Without azocasein 0.11 ± 0.05 1.2 × 108 ± 1.4 × 107 34.4 ± 2.9 0.62 ± 0.13 77.4 ± 1.63
Alpha With azocasein 0.07 ± 0.07 1.1 × 108 ± 3.1 × 107 29.5 ± 0.1 0.58 ± 0.19 77.1 ± 0.88
M. pulcherrima Without azocasein 0.08 ± 0.06 1.1 × 108 ± 1.0 × 106 28.6 ± 1.0 0.41 ± 0.11 60.4 ± 2.40
IWBT1123 With azocasein 0.07 ± 0.07 1.1 × 108 ± 4.7 × 106 26.3 ± 0.1 0.46 ± 0.09 59.8 ± 1.97
Metschnikowia spp. Without azocasein 0.08 ± 0.02 1.1 × 108 ± 3.2 × 106 24.7 ± 0.3 0.37 ± 0.03 59.5 ± 0.77
CRBO L0563 With azocasein 0.07 ± 0.02 1.0 × 108 ± 3.2 × 106 24.1 ± 0.8 0.35 ± 0.03 60.5 ± 0.77

The Kruskal–Wallis test applied to these results revealed that
azocasein had no significant impact on the growth or alcoholic fermen-
tation of the various strains tested. Statistical significance was set at
alpha = 0.05. Hence, this substrate was suitable for measuring the
proteolytic activity of the various yeast strains under oenological
conditions.
3.2. Proteolytic activity of yeasts during alcoholic fermentation
Optical density changes due to the degradation of azocasein and
growth kinetics in the various yeast cultures are shown in Fig. 1A in
Fig. 1B, respectively. No variations were observed in non-inoculated
must (data not shown) and for T. delbrueckii and S. cerevisiae cultures,
indicating the absence of protease activity, as already reported in previ-
ous study (Charoenchai et al., 1997). In contrast, a marked increase in
optical density, indicative of considerable proteolytic activity, was ob-
served for all Metschnikowia strains tested, with some variation among
strains. Under our conditions, Metschnikowia spp. CRBO L0563 exhibited
the strongest activity, leading to an increase in OD440 nm up to 0.21,
while the M. pulcherrima Y6259 sample only reached 0.17.
For all strains except M. pulcherrima Y6259, protease activity was
detectable from the beginning of the growth phase, with maximum
cumulative activity recorded at the end of growth. For M. pulcherrima
Y6259, proteolytic activity was only detected at the end of the growth
phase.
3.3. Effect of the presence of Metschnikowia on haze-forming proteins
in wine
The P3 fraction, containing the grape must proteins responsible
for haze formation, was analysed in the must and at the end of alcoholic
fermentation with S. cerevisiae, Metschnikowia spp. CRBO L0563,
M. pulcherrima IWBT Y1123, and M. pulcherrima Y6259 (Table 3). No
significant difference was observed between S. cerevisiae (55 mg eq
BSA/L) and the must not inoculated (54 mg eq BSA/L). Applying the
Kruskal–Wallis test revealed a decrease in the P3 fraction in the pres-
ence of Metschnikowia yeasts (alpha = 0.05).
4. Discussion
4.1. A new method for real-time monitoring of extracellular proteolytic
activity during alcoholic fermentation
The proteolytic activity of M. pulcherrima has already been reported
in several studies (Charoenchai et al., 1997; Comitini et al., 2011;
Fernández et al., 2000). Recently, Reid et al. (2012) identified and
sequenced the extracellular aspartic protease-encoding gene MpAPr1
from M. pulcherrima IWBT Y1123. The same kind of gene has also
been identified in another wine yeast, Candida apicola, as well as other
M. pulcherrima strains. The intensity of protease activity has been dem-
onstrated to be strain-dependent and unrelated to the gene sequence.

Fig. 1. Proteolytic activity (variations in OD440 nm) (A) and growth kinetics (B) of several yeast strains during alcoholic fermentation.
 L. Chasseriaud et al. / Journal of Microbiological Methods 119 (2015) 176–179
Table 3
Analysis of grape must proteins in the P3 fraction, obtained by HPLC, at the end of
alcoholic fermentation.
Strain P3 (mg eq BSA/L)
S. cerevisiae X5 55a ± 2
Metschnikowia spp. CRBO L0563 46b ± 2
M. pulcherrima IWBT Y1123 46b ± 2
M. pulcherrima Y6259 48b ± 3
a, b: results of Kruskal–Wallis test. Strains with different letters are significantly different
(alpha = 0.05).
For the first time, a method capable of monitoring and quantifying
the proteolytic activity of wine yeast in real-time, directly in fermenting
grape must, is reported.
When azocasein was added to the bioreactors, optical density in-
creased over time in the presence of M. pulcherrima and Metschnikowia
spp. No variation was observed in non-inoculated must (data not
shown), suggesting that the optical density variation resulted from the
yeast metabolism.
Monitoring proteolytic activity by measuring the degradation of
azocasein made it possible to determine the maximum activity period
for each strain, as well as variations over time. This method revealed
that proteolytic activity started at the beginning of growth, except in
the case of strain Y6259. Further experiments with this strain are re-
quired to explain this difference. This method highlighted inter-strain
differences that were much more difficult to detect using previously-
described techniques.
4.2. Haze-forming proteins decrease in the presence of
Metschnikowia yeast
In wine, some proteins are responsible for haze formation. Analysis
of the P3 fraction (comprising these proteins) revealed that, when
must was fermented with one of the three Metschnikowia strains
(CRBO L0563, IWBT Y1123, Y6259), the protein concentration decreased
compared to the must not inoculated and S. cerevisiae fermentations
(Table 3). Unlike S. cerevisiae, these three strains were found to secrete
a protease that hydrolysed azocasein under oenological conditions. The
decrease in the P3 fraction was probably due to the proteolytic activity
of Metschnikowia. However, the impact of this decrease (around 20%)
on protein stability has yet to be assessed.
5. Conclusion
This work devised, for the first time, a direct method for monitoring
yeast proteolytic activity in grape must throughout alcoholic fermenta-
tion. Different Metschnikowia strains exhibited variations in proteolytic
activity. Moreover, in the presence of Metschnikowia, the protein con-
centration in the P3 fraction, containing the grape proteins responsible
for haze formation, was significantly lower than with S. cerevisiae.
Nevertheless, the impact of this activity on protein stability requires
further investigation.
Acknowledgements
We kindly thank Benoit Divol (University of Stellenbosch, South
Africa) and Benoit Colonna-Ceccaldi (Pernod Ricard, France) for providing
the yeast strains.
179
References
Baleiras Couto, M.M., Reizinho, R.G., Duarte, F.L., 2005. Partial 26S rDNA restriction analysis
as a tool to characterise non-Saccharomyces yeasts present during red wine fermenta-
tions. Int. J. Food Microbiol. 102, 49–56.
Charoenchai, C., Fleet, G.H., Henschke, P.A., Todd, B.E.N., 1997. Screening of non-Saccharo-
myces wine yeasts for the presence of extracellular hydrolytic enzymes. Aust. J. Grape
Wine Res. 3, 2–8.
Combina, M., Mercado, L., Borgo, P., Elia, A., Jofre, V., Ganga, A., Martinez, C., Catania, C.,
2005. Yeasts associated to Malbec grape berries from Mendoza, Argentina. J. Appl.
Microbiol. 98, 1055–1061.
Comitini, F., Gobbi, M., Domizio, P., Romani, C., Lencioni, L., Mannazzu, I., Ciani, M., 2011.
Selected non-Saccharomyces wine yeasts in controlled multistarter fermentations
with Saccharomyces cerevisiae. Food Microbiol. 28, 873–882.
Dawes, H., Boyes, S., Keene, J., Heatherbell, D., 1994. Protein instability of wines: influence
of protein isoelectric point. Am. J. Enol. Vitic. 45, 319–326.
Dubourdieu, D., LLAUBERES, R.M., OLLIVIER, C., 1986. Estimation rapide des constituants
macromoléculaires des moûts et des vins par chromatographie liquide haute pression.
Conn. Vigne Vin. 20, 119–123.
Fernández, M., Úbeda, J., Briones, A., 2000. Typing of non-Saccharomyces yeasts with
enzymatic activities of interest in wine-making. Int. J. Food Microbiol. 59, 29–36.
Fleet, G., 1992. Spoilage yeasts. Crit. Rev. Biotechnol. 12, 1–44.
Fleet, G.H., Prakitchaiwattana, C., Beh, A., Heard, G., 2002. The Yeast Ecology of Wine
Grapes. In Biodiversity and Biotecnology of Wine Yeasts, (Kerala, India). pp. 1–17.
Hsu, J.-C., Heatherbell, D.A., 1987. Heat-unstable proteins in wine I. Characterization and
removal by bentonite fining and heat treatment. Am. J. Enol. Vitic. 38, 11–16.
Humbert-Goffard, A., 2003. Recherches sur les phénomènes enzymatiques intervenant
lors de l’élevage des vins (Bordeaux 2).
Jolly, N.P., Augustyn, O.P.H., Pretorius, I.S., 2003. The occurrence of non-Saccharomyces
cerevisiae yeast species over three vintages in four vineyards and grape musts from
four production regions of the Western Cape, South Africa. S. Afr. J. Enol. Vitic. 24,
35–42.
Lagace, L.S., Bisson, L.F., 1990. Survey of yeast acid proteases for effectiveness of wine haze
reduction. Am. J. Enol. Vitic. 41, 147–155.
Lamikanra, O., Inyang, I.D., 1988. Temperature influence on muscadine wine protein
characteristics. Am. J. Enol. Vitic. 39, 113–116.
Li, S.-S., Cheng, C., Li, Z., Chen, J.-Y., Yan, B., Han, B.-Z., Reeves, M., 2010. Yeast species
associated with wine grapes in China. Int. J. Food Microbiol. 138, 85–90.
Mateo, J.J., Maicas, S., Thieβen, C., 2015. Biotechnological characterisation of exocellular
proteases produced by enological Hanseniaspora isolates. Int. J. Food Sci. Technol.
50, 218–225.
Maturano, Y.P., Rodríguez Assaf, L.A., Toro, M.E., Nally, M.C., Vallejo, M., Castellanos de
Figueroa, L.I., Combina, M., Vazquez, F., 2012. Multi-enzyme production by pure
and mixed cultures of Saccharomyces and non-Saccharomyces yeasts during wine
fermentation. Int. J. Food Microbiol. 155, 43–50.
Mendoza, L.M., Manca de Nadra, M.C., Farías, M.E., 2007. Kinetics and metabolic behavior
of a composite culture of Kloeckera apiculata and Saccharomyces cerevisiae wine related
strains. Biotechnol. Lett. 29, 1057–1063.
Moine-Ledoux, V., Dubourdieu, D., 1999. An invertase fragment responsible for improving
the protein stability of dry white wines. J. Sci. Food Agric. 79, 537–543.
Pocock, K.F., HøJ, P.B., Adams, K.S., Kwiatkowski, M.J., Waters, E.J., 2003. Combined heat
and proteolytic enzyme treatment of white wines reduces haze forming protein con-
tent without detrimental effect. Aust. J. Grape Wine Res. 9, 56–63.
Reid, V.J., Theron, L.W., du Toit, M., Divol, B., 2012. Identification and partial char-
acterization of extracellular aspartic protease genes from Metschnikowia pulcherrima
IWBT Y1123 and Candida apicola IWBT Y1384. Appl. Environ. Microbiol. 78,
6838–6849.
Sarmento, M., Oliveira, J., Slatner, M., Boulton, R., 2000. Influence of intrinsic factors on
conventional wine protein stability tests. Food Control 11, 423–432.
Strauss, M.L.A., Jolly, N.P., Lambrechts, M.G., van Rensburg, P., 2001. Screening for the pro-
duction of extracellular hydrolytic enzymes by non-Saccharomyces wine yeasts.
J. Appl. Microbiol. 91, 182–190.
Theron, L.W., Divol, B., 2014. Microbial aspartic proteases: current and potential applica-
tions in industry. Appl. Microbiol. Biotechnol. 98, 8853–8868.
Waters, E.J., Wallace, W., Williams, P.J., 1992. Identification of heat-unstable wine proteins
and their resistance to peptidases. J. Agric. Food Chem. 40, 1514–1519.
Younes, B., Cilindre, C., Jeandet, P., Vasserot, Y., 2013. Enzymatic hydrolysis of thermo-
sensitive grape proteins by a yeast protease as revealed by a proteomic approach.
Food Res. Int. 54, 1298–1301.
Zott, K., Miot-Sertier, C., Claisse, O., Lonvaud-Funel, A., Masneuf-Pomarede, I., 2008.
Dynamics and diversity of non-Saccharomyces yeasts during the early stages in
winemaking. Int. J. Food Microbiol. 125, 197–203.