Vicinal diketones and their precursors in wine alcoholic fermentation: Quantification and dynamics of production

Vicinal diketones and their precursors in wine alcoholic fermentation: Quantification and dynamics of production

Accepted Manuscript Vicinal diketones and their precursors in wine alcoholic fermentation: Quantification and dynamics of production Thomas Ochando, ...

2MB Sizes 1 Downloads 64 Views

Accepted Manuscript Vicinal diketones and their precursors in wine alcoholic fermentation: Quantification and dynamics of production

Thomas Ochando, Jean-Roch Mouret, Anne Humbert-Goffard, Jean-Marie Sablayrolles, Vincent Farines PII: DOI: Reference:

S0963-9969(17)30724-X doi:10.1016/j.foodres.2017.10.040 FRIN 7088

To appear in:

Food Research International

Received date: Revised date: Accepted date:

6 July 2017 26 September 2017 19 October 2017

Please cite this article as: Thomas Ochando, Jean-Roch Mouret, Anne Humbert-Goffard, Jean-Marie Sablayrolles, Vincent Farines , Vicinal diketones and their precursors in wine alcoholic fermentation: Quantification and dynamics of production. The address for the corresponding author was captured as affiliation for all authors. Please check if appropriate. Frin(2017), doi:10.1016/j.foodres.2017.10.040

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Vicinal diketones and their precursors in wine alcoholic fermentation: quantification and dynamics of production

PT

Thomas Ochando1,2, Jean-Roch Mouret1, Anne Humbert-Goffard2, Jean-

SC

RI

Marie Sablayrolles1, Vincent Farines1

1

Moët & Chandon, F-51200 Epernay, France

MA

2

NU

UMR SPO: INRA, Université Montpellier, Montpellier SupAgro, 34060, Montpellier, France

*Corresponding author: Tel: +33-4-99-61-22-74; Fax: +33-4-99-61-28-51

AC

CE

PT E

D

E-mail address: [email protected]

1

ACCEPTED MANUSCRIPT Abstract Vicinal diketones produced during wine fermentation influence the organoleptic qualities of wine. Diacetyl and 2,3-pentanedione are well known for their contribution to butter or butterscotch-like flavours. We developed an analysis method to quantify vicinal diketones and their precursors, α-acetolactate and α-acetohydroxybutyrate, under oenological

PT

conditions. Five-fold dilution of the sample in a phosphate-citrate buffer (pH 7.0) strongly attenuated matrix effects between the beginning and end of alcoholic fermentation and

RI

protected the sample from spontaneous precursor decarboxylation. The use of diacetyl-d6 as

SC

an internal reference improved precision by eliminating differences in the derivatization and extraction yields between the internal standard and the analytes. We obtained unexpected

NU

results for alcoholic fermentation by Saccharomyces cerevisiae using this approach. Indeed, the level of diacetyl and 2,3-pentanedione throughout fermentation were very low. However,

MA

we observed a large quantity of both precursors. The production dynamics of α-acetolactate were unconventional and there were two distinct phases of accumulation. The first

D

corresponded to the growth phase, and the second to glucose depletion. There was a rapid

PT E

decrease of precursor levels at the end of fermentation, but there was still a significant amount of α-acetolactate. The amount of precursor remaining at the end of fermentation

CE

constitutes a potential source of diacetyl during wine maturation. α-acetohydroxybutyrate accumulated during the growth phase followed by a continuous decrease of its concentration

AC

during the stationary phase. Residual quantities of α-acetohydroxybutyrate found in wine at the end of fermentation does not constitute a sufficient source of 2,3-pentanedione to affect the aromatic profile.

Keywords Wine fermentation, Diacetyl, Pentanedione, α-Acetolactate, α-Acetohydroxybutyrate, Production, Quantification

2

ACCEPTED MANUSCRIPT 1. Introduction Diacetyl (2,3-butanedione) and 2,3-pentanedione are vicinal diketones well known for their contribution to butter or butterscotch-like flavours in food matrices (van Neil, Kluyver, & Derx, 1929). Diacetyl is found in fermented products such as cheese, beer, and wine. Diacetyl and 2,3-pentanedione arise from microbial metabolism of yeast and bacteria (Chuang & Collins,

PT

1968). Diacetyl in wine is often related to malolactic fermentation. Thus, up to now, the focus of most of publications has been directed towards interactions between yeast and bacteria

RI

such as Streptococcus, Leuconostoc, Lactobacillus, Pediococcus, and Oenococcus

SC

(Bartowsky & Henschke, 2004; Chuang & Collins, 1968, 1972; Martineau, Acree, & HenickKling, 1995; Mascarenhas, 1984; Renouf & Murat, 2010; Suomalainen & Ronkainen, 1968).

NU

These numerous researches on lactic bacteria have leaded to the selection of citrate lyase negative starter cultures which lowly degrade citric acid providing a lower production of

MA

buttered notes. Brewers have focused their research on yeast alone because no malolactic fermentation occurs during malt fermentation (Duong et al., 2011; Erten, Tanguler, &

D

Cakiroz, 2007; García, García, & Díaz, 1994; Gibson et al., 2014; Inoue, Masuyama,

PT E

Yamamoto, & Okada, 1968; Jones, Margaritis, & Stewart, 2007; Krogerus & Gibson, 2015, 2013a; Martin Marais, 2010; Sheppard, 2007). Finally, few publications have addressed the

CE

production of diacetyl by yeast alone in winemaking conditions. The malolactic fermentation is not systematic in the wine production; therefore, it is important to characterize the

AC

dynamics of production of vicinal diketones and their precursors by yeasts during alcoholic fermentation. Moreover, the quantity and the impact of precursors at the end of fermentation is not fully understood. Diacetyl is produced from α-acetolactate by chemical non-enzymatic oxidative decarboxylation, whereas 2,3-pentanedione is formed from α-acetohydroxybutyrate from the same chemical mechanism (de Man, 1959; Krogerus & Gibson, 2013a; Monnet et al., 2003; Suomalainen & Ronkainen, 1968; White & Wainwright, 1975). These two αacetohydroxyacids, which are intermediates in the synthesis pathways of valine, leucine, and 3

ACCEPTED MANUSCRIPT isoleucine, are produced in the cells, and then excreted in the must where they are chemically converted into vicinal diketones. Finally, diacetyl and 2,3-pentanedione can be enzymatically reduced by yeast into acetoin and 3-hydroxy-2-pentanone, which are also converted into 2,3-butanediol and 2,3-pentanediol. The metabolic pathways are described in figure 1.

PT

From a sensory point of view, perception threshold of diacetyl in wine depends on the type of wine (Martineau et al., 1995). Matrices of white wines are less complex than those of red

RI

wines, notably due to low polyphenol contents. For example, the sensory threshold of

SC

diacetyl in chardonnay is approximately 0.2 mg/L. In contrast, the threshold in red wines, such as cabernet sauvignon, is approximately 2.8 mg/L (Martineau et al., 1995). The flavour

NU

threshold of 2,3-pentanedione is approximately 0.9 mg/L, but the 2,3-pentanedione content

MA

in fermented beverages is lower than 0.1 mg/L (Meilgaard, 1975). Thus, 2,3-pentanedione has a minor impact on the flavour of fermented beverages. Acetoin and 2,3-butanediol can be found in dry wines, with average concentrations of 5-20 mg/L and 0.2-0.7 g/L respectively

D

(P. Romano, Brandolini, Ansaloni, & Menziani, 1998; Patrizia Romano & Suzzi, 1996).

PT E

However, neither 2,3-butanediol nor acetoin are strongly odorous; in fact, their threshold values in wine are very high, both being higher than 150 mg/liter (Dubois, 1994). Quantities

CE

of 3-hydroxy-2-pentanone and 2,3-pentanediol in wines are very much lower than acetoin and 2,3-butanediol, their aromatic contribution is generally considered as very limited

AC

(Ronkainen, Brummer, & Suomalainen, 1970). Analytical aspects are important to study the dynamics of diketones production linked to acetohydroxyacid precursors formation during the fermentation. Quantification of these molecular markers during fermentation is essential but complex because of their instability due to temperature and redox potential (Inoue et al., 1968). Moreover, substantial modifications of the matrix with the evolution of the sugar/ethanol ratio can interfere with assay method. Some methodologies have been described in the literature: a fluorometric method with Rhodamine B Hydrazide (RBH) (Li, Duerkop, & Wolfbeis, 2009), a colorimetric 4

ACCEPTED MANUSCRIPT method with creatine and α-naphthol (Mattessich & Cooper, 1989), and liquid or gas chromatography using quinoxaline derivatives (Landaud, Lieben, & Picque, 1998; Martineau, Acree, & Henick-Kling, 1994; Otsuka & Ohmori, 1992; Revel, Pripis-Nicolau, Barbe, & Bertrand, 2000). Colorimetric and fluorometric methods are not selective and many compounds can interfere. In contrast, chromatographic methods coupled to mass

PT

spectrometry are selective and allow reliable quantitative determination (Landaud et al., 1998; Martineau et al., 1994).

RI

Finally, free vicinal diketones quantification is important in relation to the quality of the wine

SC

but the determination of precursors is also a crucial issue to predict the potential aromatic impact by their decarboxylation during the aging process. In this sense, this work (i) provides

NU

a substantial improvement of an existing gas chromatography coupled to mass spectrometry method for the quantification of diacetyl, 2,3-pentanedione and their precursors (α-

MA

acetolactate and α-acetohydroxybutyrate) during alcoholic fermentation and (ii) describes the

PT E

D

production dynamics of both compounds in natural and synthetic musts.

2. Materials and methods

Quantification of diacetyl, 2,3-pentanedione and their precursors (α-

CE

2.1.

acetolactate and α-acetohydroxybutyrate)

2.1.1.1.

AC

2.1.1. Sample preparation Sampling

Before sampling, 4 mL inerted phosphate-citrate buffer, pH 7, containing 1 mg/L deuterated diacetyl (2,3-butanedione-d6, CAS 22026-37-5, CDN Isotopes, Quebec, Canada) was pipetted into a 20-mL vial and the headspace inerted with argon for a few seconds, before closing the vial. During alcoholic fermentation, 7 mL of fermenting medium were sampled and centrifuged at 3500  g at 4°C for 10 min. The 7mL was splitted into two parts. First, 1 5

ACCEPTED MANUSCRIPT mL of supernatant was added with a Gastight® syringe through the septum of an inerted vial containing the phosphate-citrate buffer (Figure 2). Then, this first vial was stored at -20°C until the analysis of vicinal diketones (diacetyl and 2,3-pentanedione). Same sampling preparation in a second vial was done for determination of the total potential vicinal diketones (i.e. including precursors) content. Derivatization

PT

2.1.1.2.

RI

For the determination of vicinal diketones, 0.5 mL of 20 mM 4,5-Dichloro-O-

SC

phenylenediamine (DCDB, CAS 5348-42-5, Sigma-Aldrich, Saint-Quentin Fallavier, France) in 1 M HCl was added to the vial. The reaction mixture was maintained at 30°C for 5 min

NU

(Figure 2). The result of the derivatization is 6,7-dichloro-2,3-dimethylquinoxaline (DCDMQ) for diacetyl and 6,7-dichloro-2-methyl-3-ethylquinoxaline (DCMEQ) for 2,3-pentanedione. Oxidation step

MA

2.1.1.3.

D

The determination of total diacetyl and pentanedione content (including α-acetolactate and α-

PT E

acetohydroxybutyrate precursors, respectively) was carried out by oxidative decarboxylation of α-hydroxyacids in the presence of a transition metal (Paria, Chatterjee, & Paine, 2014). Seventy-five microliters of a 10 mM FeSO4 and FeCl3 solution was added to the sample and

CE

the sample shaken vigorously in the presence of air. Oxidation was performed at 80°C for 10 min. After cooling, the sample was derivatized (Figure 2). Oxidation conditions (temperature

AC

and time) were previously optimized using a synthetic -acetolactate solution (data not shown). The oxidation yield was between 99 and 100% under these conditions. 2.1.1.4.

Extraction step

The extraction of DCDMQ and DCMEQ was performed by adding 1.5 ml anhydrous toluene (ChromasolvR Plus,  99.9%, CAS 108-88-3, Sigma-Aldrich, Saint-Quentin Fallavier, France) containing 500 µg/L anhydrous dodecane ( 99.9%, CAS 112-40-3, Sigma-Aldrich, SaintQuentin Fallavier, France). Dodecane was initially used as an internal standard for method 6

ACCEPTED MANUSCRIPT development. Extraction was performed by slowly rolling the vials on a tilting table for 120 seconds to obtain reproducible results. Next, 1 mL organic phase was removed and transferred into a 1.5 mL vial. One microliter was injected in splitless mode for analysis by GC-MS.

PT

2.1.2. GS/MS apparatus and analytical conditions Analysis were performed using a TRACE™ Ultra Gas Chromatograph coupled to an ISQTM

RI

series (single quadrupole) Mass Spectrometer detector (ThermoFisher Scientific™, Villebon

SC

sur Yvette, France). Separation of the analytes was carried out with a DB-5MS methyl siloxane column (30 m x 0.25 mm, 0.25 µm film thickness) from Agilent Technologies

NU

(Courtaboeuf, France) using helium as a carrier gas (flow rate: 1.0 ml/min). The following GC temperature program was used: injector temperature 250°C; initial oven temperature: 70°C;

MA

initial holding time: 1.5 min; temperature increment rate: 25°C/min up to 200°C, followed by 10°C/min up to 270°C. The retention times observed for DCDMQd6, DCDMQ, and DCMEQ

D

under the separation conditions used were 8.76, 8.79, and 9.39 min, respectively. Detection

PT E

was performed in Selective Ion Monitoring (SIM) mode. The MS source temperature was maintained at 250°C. Qualification of DCDMQd6, DCDMQ, and DCMEQ was based on the following m/z ions: 74, 109, 144, 185, 188, 226, 232, and 240 monitored for retention times

CE

between 6 and 13 min. Selective m/z ions 232, 226, and 240 were used for quantitation of

AC

DCDMQd6, DCDMQ, and DCMEQ, respectively. Data were processed on Xcalibur™ Software from Thermo Fisher Scientific. Quantification was performed by comparison to calibration curves obtained from the analysis of known amounts of diacetyl and added to synthetic media (must to wine). Ratios of the DCDMQ or DCMEQ peak areas to the internal deuterated standard were used for calibration and quantitation.

2.2.

Media and Fermentation

7

ACCEPTED MANUSCRIPT 2.2.1. Media Synthetic media for analysis

The method for diacetyl and 2,3-pentanedione quantification was tested in various synthetic media simulating initial must, musts at various stages of fermentation, and wine. For the model solution simulating must, 180 g/L total sugars (glucose/fructose 1:1) were added to a

PT

buffered solution containing 7 g/L tartaric acid and 7 g/L malic acid adjusted to pH 3.1 with

RI

sodium hydroxide. For the model solution simulating wine, the buffer solution was supplemented with 84.6 g/L ethanol, without the addition of sugars. Model solutions of musts

SC

at various stages of fermentation were obtained by supplementing the buffer solution

NU

described above with (1) 135 g/L sugars and 21.2 g/L ethanol, (2) 90 g/L sugars and 42.3 g/L ethanol, or (3) 45 g/L sugars and 63.5 g/L ethanol. The resulting media corresponded to 15,

MA

50, and 75% fermentation. Synthetic must for fermentation

D

Some fermentations were performed in a synthetic medium (SM) that mimicked Champagne

PT E

must (Ochando, Mouret, Humbert-Goffard, Sablayrolles, & Farines, 2016) and derived from standard grape juice, as described by Bely (Bely et al. 1990). This culture medium contained

CE

180 g/L sugars (glucose/fructose 1:1), 7 g/L malic acid, 7 g/L tartric acid, salts (0.75 g/L KH2PO4, 0.50 g/L K2SO4, 0.25 g/L MgSO4, 0.155 g/L CaCl2, and 0.20 g/L NaCl), vitamins (20

AC

mg/L myo-inositol, 1.5 mg/L pantothenic acid, 0.25 mg/L thiamine, 2 mg/L nicotinic acid, 0.25 mg/L pyridoxine, and 0.003 mg/L biotin), and trace elements (4 mg/L MnSO4, 4 mg/L ZnSO4, 1 mg/L CuSO4, 1 mg/L KI, 0.4 mg/L CoCl2, 1 mg/L H3BO3, and 1 mg/L (NH4)6Mo7O24). The pH of the medium was adjusted to 3.1 with 10 M NaOH. The source of nitrogen was a mixture of ammonium (30%) and amino acids (70%), as commonly found in pinot noir must. The assimilable nitrogen concentration was 360 mg/L. The composition of the amino acid solution was as follows (in mg/L): tyrosine (8.3), tryptophan (6.5), isoleucine (15.9), aspartate (15.1), glutamate (56.9), arginine (550.3), 8

ACCEPTED MANUSCRIPT leucine (20.5), threonine (124.3), glycine (3.5), glutamine (180), alanine (221.3), valine (30.9), methionine (5.9), phenylalanine (20.8), serine (74.7), histidine (12.6), lysine (4.6), cysteine (15.1), and proline (100.5). This solution was established based on an assay of natural Champagne grape musts from pinot noir. To obtain 360 mg/L assimilable nitrogen in the SM, 14.6 mL of an amino-acid stock solution and 413 mg of NH4Cl were added to 1 litre

PT

of medium. The SM medium was initially supplemented with lipids, a mix of phytosterols (CAS: 85.451,

RI

Sigma Aldrich, Saint Quentin Fallavier, France) and fatty acids. The stock solution was

SC

composed of 15 g/L β-sitosterol in Tween 80 and absolute ethanol (1:1, v/v). To obtain a concentration of 2 mg/L initial phytosterols, 1 mL/L of a10-fold-diluted stock solution was

NU

added to the medium. Total desorption of oxygen was monitored using a PreSens® oxygen

MA

probe (PreSens Precision Sensing GmbH, Regensburg, Germany). Natural must

D

Various natural musts of pinot noir from Champagne were used. The musts were from

PT E

different vintages, and the levels of yeast available nitrogen (YAN), ammonium, and pH were different. Sugars were standardized by chaptalization to 180 -185 sugars g/L. Each fresh

CE

must was analyzed at the time of harvest. The musts characteristics are described in table 1.

AC

2.2.2. Fermentation

Synthetic must Fermentations of the synthetic musts were made in replicates in 10 L stainless steel fermenters, with 9 L must, at 20°C. The amount of CO2 released was automatically measured with a gas mass flow meter for online estimation of the rate of CO2 production (dCO2/dt). Fermentation was carried out with the Saccharomyces cerevisiae strain isolated from the Champagne vineyard, [28]. Fermentation tanks were inoculated with 10 g/hL active dry yeast previously rehydrated for 30 min at 30 °C in a 50 g/L glucose solution.

9

ACCEPTED MANUSCRIPT Natural musts Fermentation of natural musts was performed at an industrial scale in various tanks of between 6 and 120 hL. Fermentation was carried out using the same Saccharomyces cerevisiae strain at 10 g/hL of active dry yeast. The fermentation volumes were too large to ensure the implementation of replicates.

Development of the analytical method

RI

3.1.

PT

3. Results and discussion

We calibrated the assay using concentrations from 0.025 to 2 mg/L diacetyl and 2,3-

SC

pentanedione in various synthetic media that stimulated the progress of fermentation

NU

between 0 and 100% to study the matrix effect on quantification. These standard solutions were analysed with the method described by Landaud et al., 1998 using dodecane as

MA

internal standard. The slope of the standard curves varied with the composition of the matrix during alcoholic fermentation, (Figure 3). The difference of the measurement between must (0%) and wine (100%) was 2.6 and 2-fold for diacetyl and 2,3-pentanedione, respectively.

D

Also, the standard curves were not linear beyond 1 mg/L of the two compounds. These

PT E

results showed that this method was not adapted to oenological conditions, including the study of diacetyl and its precursor in a matrix composition, which strongly changes during

CE

alcoholic fermentation. Indeed, the high levels of sugars contained in the must and the gradual formation of ethanol during fermentation probably modify the yield of the extraction of

AC

the molecules (post-derivatization) by toluene. It was thus necessary to alter the method to decrease matrix effects and increase the range of linearity. Diluting the sample decreased the co-extractive matrix effects (i.e. sugars and ethanol), but also decreased the concentration of the target analytes. Higher dilutions further reduced or eliminated the matrix effects, but also necessitated higher sensitivity as dilution factors increased, requiring a compromise. Dilutions from 5 to 100-fold were tested. Choosing the proper dilution factor depends on the concentrations of analytes and coextractives, as well as chemical properties and instrument sensitivity. The excellent 10

ACCEPTED MANUSCRIPT sensitivity of mass spectrometry in the SIM mode can accommodate the inherent sensitivity challenge of dilution-based methods. We chose to dilute the sample 1:5 in a phosphatecitrate buffer at pH 7 to drastically decrease the matrix effects. A five-fold dilution (i) decreases the difference between sugar and ethanol concentrations during alcoholic fermentation and (ii) neutralizes sample pH, limiting the risk of spontaneous decarboxylation

PT

of the precursors. Inert vials containing 4 mL phosphate citrate buffer at pH 7 can be prepared in advance and stored at 4°C before sampling. During fermentation, 1 mL of

RI

centrifuged supernatant can be added to the vial and the vial stored at -20°C until analysis.

SC

Table 2 presents the calibrations results for diacetyl and 2,3-pentanedione in media mimicking 0, 50, and 100% fermentation diluted five-fold in phosphate-citrate buffer at pH 7.

NU

At this dilution, the matrix effect was limited and the curves were linear until 1 mg/L and the

MA

slopes for the different media were very close. For example mean slope for diacetyl was of 0.508 with a CV% of 4.41.

D

The last step of adaptation of the assay method to oenological conditions concerned the

PT E

internal standard to be used. Dodecane was used as the internal standard in the study of Landaud et al., 1998. Dodecane was directly added in the organic phase (toluene) before the extraction step, whereas diacetyl and 2,3-pentanedione were in the aqueous phase of the

CE

sample. Thus, the yields of diacetyl and 2,3-pentanedione were subject to the influence of the derivatization and extraction steps, whereas that of dodecane was not. We substantially

AC

modified the approach by changing the internal standard, and the phase to which it is added, to improve the accuracy and precision of the method. The proper internal standard should be chemically similar to the analysed compound but not naturally present in the sample. Furthermore, it should have the same functional group, boiling point, and activity as the target molecule. We chose to use Deuterated diacetyl (diacetyl-d6) because it is suitable for MS detection. We directly added diacetyl-d6 to the phosphate-citrate buffer. Thus, diacetyl-d6 underwent the same preparation steps as the target molecules, diacetyl and pentanedione, with the same derivatization and extraction yields. The mass lag between DCDMQ and 11

ACCEPTED MANUSCRIPT DCDMQ-d6 allowed identification of both compounds. The ions observed were 226,185, 144, 109, 74 for DCDMQ and 232,188, 144, 109, 74 for DCDMQ-d6 and 240, 226, 185, 144, 109, 74 for DCMEQ. We tested diacetyl-d6 as an internal standard at 0, 50, and 100% fermentation at concentrations between 0.025 and 2 mg/L of diacetyl and 2,3-pentanedione. Calibration

PT

results are presented in table 2. The range of linearity, until 2 mg/L of diacetyl, was larger than with only dodecane. The mean slope between the three different fermentation

RI

progresses was 0.3274 with a CV% of 0.87 for diacetyl and 0.3381 with a CV% 1.25 for 2,3-

SC

pentanedione. Thus, the combination of a five-fold dilution in phosphate-citrate buffer and the use of deuterated internal standard in this phase allow a very accurate quantification vicinal

NU

diketones during the whole alcoholic fermentation.

MA

The modified method makes it possible to measure the production of diketones and their precursors. Indeed, it is possible to quantify (1) the content of diketones (diacetyl and 2,3-

D

pentanedione), (2) total diketones (sum of concentrations of diacetyl, 2,3-pentanedione and

PT E

their precursors) and to calculate the quantity of alpha-acetohydroxyacid precursors, expressed as diacetyl and 2,3-pentanedione equivalents. Such expression of precursors as equivalents is informative because these molecules constitute the potential source of diacetyl

3.2.

AC

Figure 2.

CE

and 2,3-pentanedione. The steps of the assay method are summarized in the flowchart of

Production of diacetyl, 2,3-pentanedione and precursors during alcoholic fermentation of wine by S. cerevisiae

3.2.1. Fermentation in synthetic must We first determined the production dynamics of diacetyl, 2,3-pentanedione and their precursors under oenological conditions by multi-sampling during alcoholic fermentation in synthetic must at 20°C. Samples for the determination of vicinal diketones and their precursors were taken every six hours during fermentation. There was residual assimilable 12

ACCEPTED MANUSCRIPT nitrogen at a concentration of 117 mg/L at the end of the growth phase, approximately 50 hours (Figure 4a), indicating that fermentation was limited by lipids, as previously shown by Ochando et al., 2016. 3.2.1.1.

Production dynamics of diacetyl and its precursor α-acetolactate

PT

We determined the concentrations of diacetyl and α-acetolactate, expressed in diacetyl equivalents (eq.), by determining the difference between total diacetyl (oxidized sample) and

RI

diacetyl (inert sample) (Figure 4b).

SC

There was a first phase of production during the growth phase, corresponding to the first 50 hours. At the peak of the first production phase, α-acetolactate represented 96% of total

NU

diacetyl. This peak corresponded to the maximum CO2 production rate (Figure 4a), i.e. the end of the growth phase. During the first 100 hours of fermentation, total diacetyl consisted

MA

almost exclusively of α-acetolactate, with very little diacetyl content. There was then a second phase of α-acetolactate accumulation in the medium starting after 100 hours (at

D

approximately 60% fermentation), resulting in a second peak at 86% fermentation. The

PT E

concentration then decreased until the end of fermentation. At the second peak, diacetyl represented 35% and its precursor 65%.

CE

The first peak of α-acetolactate production was due to the metabolic switch between the growth and stationary phases. This metabolic switch induced a lower carbon flux of

AC

glycolysis. Thus, α-acetolactate was consumed more rapidly than it was produced. The origin of the second peak is less clear. The second peak corresponded to glucose depletion (< 5 g/L). Glucose induces Pyruvate Decarboxylase (PDC), (Kellermann & Hollenberg, 1988; Schmitt, Ciriacy, & Zimmermann, 1983). The second peak of α-acetolactate may be due to a change in the balance of the pyruvate flux between PDC and acetolactate synthase linked to the level of glucose and ethanol in the medium. We observed a peak of free diacetyl concomitant to the second peak of α-acetolactate, corresponding to spontaneous decarboxylation of the precursor in the medium. This

13

ACCEPTED MANUSCRIPT physicochemical reaction is favoured by a low pH 3.1 and the higher temperature than that of brewery conditions, (Krogerus & Gibson, 2013a).

Production dynamics of 2,3-pentanedione and its precursor αacetohydroxybutyrate

PT

3.2.1.2.

We also monitored 2,3-Pentanedione and its precursor during alcoholic fermentation (Figure

RI

6c). The level of pentanedione was very low and α-acetohydroxybutyrate was the major

SC

compound present during alcoholic fermentation. The production dynamics of 2,3pentanedione and its precursor were very similar until the end of the growth phase, unlike

NU

that of diacetyl and -acetolactate. At the maximum of the only production peak (at 50 h), αacetohydroxybutyrate represented 66% of total 2,3-pentanedione. The concentration of α-

MA

acetohydroxybutyrate then decreased until the end of fermentation. This profile is expected, as 2,3-pentanedione is linked to isoleucine production from threonine degradation, (Krogerus

D

& Gibson, 2013b). 2,3-pentanedione production is thus related to nitrogen metabolism

PT E

(Figure 1). At the end of the growth phase, nitrogen consumption stopped. Thus production of α-acetohydroxybutyrate was lower than its conversion to 2,3-dihydroxy-3-methylvalerate.

CE

Monitoring the production of diacetyl, 2,3-pentanedione and its precursors by yeast, during alcoholic fermentation, revealed unexpected dynamics of diacetyl and α-acetolactate

AC

production. Diacetyl levels during alcoholic fermentation were very low, whereas levels of the precursor varied widely. Indeed, biphasic accumulation of α-acetolactate in must during alcoholic fermentation has not been previously reported. We monitored these molecules during alcoholic fermentation in natural must to determine the generalisability of this observation. 3.2.2. Fermentation in natural must Production of vicinal diketones and their precursors was monitored during alcoholic fermentation of a pinot noir (n°1) described in $2.2.1. The global results were very similar to 14

ACCEPTED MANUSCRIPT those observed in synthetic must, even though the total quantity of vicinal diketones and their precursors were higher, resulting in higher residual levels at the end of fermentation (Figure 5). The first phase of α-acetolactate production during the growth phase was followed by a second phase of α-acetolactate accumulation in the medium, with a maximum value after

PT

approximately 70% of fermentation, at 120 h approximately (figure 5a). Quantities of αacetohydroxybutyrate was higher than 2,3-pentanedione throughout alcoholic fermentation,

SC

RI

and its production occurring during the first third of the process (figure 5b).

We have then compared the production of total diketone in three different musts of pinot noir

NU

at 20°C (Figure 6). The parameters of the musts are described in the materials and methods, $2.3.1. In each case, we obtained two phases of production for total diacetyl. Surprisingly,

MA

there was no correlation between the accumulated level of α-acetolactate at 70 – 80 % of fermentation and its final residual value (figure 6). Indeed, the final concentration of total

D

diacetyl was almost identical for the three musts of pinot noir, whereas the corresponding

4. Conclusion

PT E

concentration at 80% of fermentation ranged from 0.9 to 1.4 mg/L.

CE

We modified a method for GC/MS quantification of diacetyl, 2,3-pentanedione and its precursors, α-acetolactate and α-acetohydroxybutyrate, to evaluate the concentration of

AC

these molecules under oenological conditions throughout the alcoholic fermentation process. Five-fold dilution of the sample supernatant taken during fermentation eliminated matrix effects and protected the sample from spontaneous precursor decarboxylation. The use of a deuterated internal standard also made the method more precise by eliminating differences in the derivatization and extraction yields between the internal standard and the analytes. This method was applied to sampling during alcoholic fermentation in synthetic and natural musts. The results show the presence of very low concentrations of diacetyl and 2,3pentanedione during the process. Residual concentrations of diacetyl were null or below 50

15

ACCEPTED MANUSCRIPT µg/L at the end of the process. The residual concentrations of 2,3-pentanedione were even lower, close to zero. However, there were high concentrations of their precursors during alcoholic fermentation. α-acetolactate production was highlighted by two phases of accumulation. The first coincided with the growth phase of the yeast and the second ended between 80 and 90% completion of fermentation. Consequent to the second production

PT

phase, there was a rapid decrease in the precursor concentration. At the end of fermentation, we observed various quantities of residual α-acetolactate that were independent of the

RI

concentration at the second peak.

SC

The quantification of diacetyl precursors under oenological conditions is necessary because α-acetolactate decarboxylation can spontaneously occur in the presence of oxygen. Thus, α-

NU

acetolactate represents a potential reservoir of diacetyl and must be considered, because

MA

butter or butterscotch-like flavours can appear during storage of the wine and be detected at the time of wine tasting, even if no diacetyl was detected at the end of alcoholic fermentation. α-acetohydroxybutyrate was also produced during alcoholic fermentation, but there was only

D

one phase of production, occurring during early fermentation, i.e. the growth phase. The

PT E

residual quantities found in wine were below the threshold value. Here, we determined the production dynamics of vicinal diketones and their precursors

CE

during alcoholic fermentation. The next step will be to determine the influence of the main

AC

parameters on the production of these compounds.

16

ACCEPTED MANUSCRIPT Bibliography Bartowsky, E. J., & Henschke, P. A. (2004). The ‘buttery’ attribute of wine—diacetyl—desirability, spoilage and beyond. International Journal of Food Microbiology, 96(3), 235‑ 252. https://doi.org/10.1016/j.ijfoodmicro.2004.05.013 Chuang, L. F., & Collins, E. B. (1968). Biosynthesis of diacetyl in bacteria and yeast. Journal of

PT

Bacteriology, 95(6), 2083‑ 2089. Chuang, L. F., & Collins, E. B. (1972). Inhibition of Diacetyl Synthesis by Valine and the Roles of -

RI

Ketoisovaleric Acid in the Synthesis of Diacetyl by Saccharomyces cerevisiae. Journal of

SC

General Microbiology, 72(1), 201‑ 210. https://doi.org/10.1099/00221287-72-1-201 de Man, J. C. (1959). The formation of diacetyl and acetoin from α-acetolactic acid. Recueil Des

NU

Travaux Chimiques Des Pays-Bas, 78(7), 480‑ 486. https://doi.org/10.1002/recl.19590780703 Dubois, P. (1994). Les arômes des vins et leurs défauts. Revue française d’oenologie, 34(146), 39‑ 50.

MA

Duong, C. T., Strack, L., Futschik, M., Katou, Y., Nakao, Y., Fujimura, T., … Nevoigt, E. (2011). Identification of Sc-type ILV6 as a target to reduce diacetyl formation in lager brewers’ yeast.

D

Metabolic Engineering, 13(6), 638‑ 647. https://doi.org/10.1016/j.ymben.2011.07.005

PT E

Erten, H., Tanguler, H., & Cakiroz, H. (2007). The effect of pitching rate on fermentation and flavour compounds in high gravity brewing. Journal of the Institute of Brewing, 113(1), 75–79.

CE

García, A. I., García, L. A., & Díaz, M. (1994). Modelling of diacetyl production during beer fermentation. Journal of the Institute of Brewing, 100(3), 179–183.

AC

Gibson, B., Krogerus, K., Ekberg, J., Monroux, A., Mattinen, L., Rautio, J., & Vidgren, V. (2014). Variation in α -acetolactate production within the hybrid lager yeast group Saccharomyces pastorianus and affirmation of the central role of the ILV6 gene: α -Acetolactate production by lager yeast. Yeast, n/a-n/a. https://doi.org/10.1002/yea.3026 Inoue, T., Masuyama, K., Yamamoto, Y., & Okada, O. (1968). Mechanism of diacetyl formation in beer.III. Proceeding of the American Society of Brewing Chemists, 158, 17‑ 23.

17

ACCEPTED MANUSCRIPT Jones, H. L., Margaritis, A., & Stewart, R. J. (2007). The Combined Effects of Oxygen Supply Strategy, Inoculum Size and Temperature Profile on Very-High-Gravity Beer Fermentation by Saccharomyces cerevisiae. Journal of the Institute of Brewing, 113(2), 168–184. Kellermann, E., & Hollenberg, C. P. (1988). The glucose-and ethanol-dependent regulation of PDC1 from Saccharomyces cerevisiae are controlled by two distinct promoter regions. Current

PT

Genetics, 14(4), 337‑ 344. Krogerus, & Gibson. (2015). An Improved Model for Prediction of Wort Fermentation Progress and

SC

https://doi.org/10.1094/ASBCJ-2015-0106-01

RI

Total Diacetyl Profile. Journal of the American Society of Brewing Chemists.

NU

Krogerus, K., & Gibson, B. R. (2013a). 125 th Anniversary Review: Diacetyl and its control during brewery fermentation: Diacetyl and its control during brewery fermentation. Journal of the

MA

Institute of Brewing, n/a-n/a. https://doi.org/10.1002/jib.84 Krogerus, K., & Gibson, B. R. (2013b). Influence of valine and other amino acids on total diacetyl and

D

2,3-pentanedione levels during fermentation of brewer’s wort. Applied Microbiology and

PT E

Biotechnology, 97(15), 6919‑ 6930. https://doi.org/10.1007/s00253-013-4955-1 Landaud, S., Lieben, P., & Picque, D. (1998). Quantitative analysis of diacetyl, pentanedione and their precursors during beer fermentation by an accurate GC/MS method. Journal of the Institute

CE

of Brewing, 104(2), 93–99.

AC

Li, X., Duerkop, A., & Wolfbeis, O. S. (2009). A Fluorescent Probe for Diacetyl Detection. Journal of Fluorescence, 19(4), 601‑ 606. https://doi.org/10.1007/s10895-008-0450-y Martin Marais, L. (2010). The effect of yeast propagation temperature on diacetyl reduction. An inprocess study at Spendrups brewery. Consulté à l’adresse http://publications.lib.chalmers.se/records/fulltext/163053.pdf Martineau, B., Acree, T. E., & Henick-Kling, T. (1995). Effect of wine type on the detection threshold for diacetyl. Food Research International, 28(2), 139‑ 143. https://doi.org/10.1016/09639969(95)90797-E 18

ACCEPTED MANUSCRIPT Martineau, B., Acree, T., & Henick-Kling, T. (1994). A simple and accurate GC/MS method for the quantitative analysis of diacetyl in beer and wine. Biotechnology techniques, 8(1), 7–12. Mascarenhas, M. A. (1984). The Occurrence of Malolactic Fermentation and Diacetyl Content of Dry Table Wines from Northeastern Portugal. American Journal of Enology and Viticulture, 35(1), 49.

PT

Mattessich, J., & Cooper, J. R. (1989). The spectrophotometric determination of diacetyl. Analytical Biochemistry, 180(2), 349‑ 350. https://doi.org/10.1016/0003-2697(89)90443-0

RI

Meilgaard, M. . (1975). Flavor chemistry of beer: part ii: flavour and threshold of 239 aroma volatiles.

SC

Master Brewers Association of the Amercias, 12(3), 151‑ 168.

NU

Monnet, C., Nardi, M., Hols, P., Gulea, M., Corrieu, G., & Monnet, V. (2003). Regulation of branchedchain amino acid biosynthesis by alpha-acetolactate decarboxylase in Streptococcus

MA

thermophilus. Letters in Applied Microbiology, 36(6), 399‑ 405. Ochando, T., Mouret, J.-R., Humbert-Goffard, A., Sablayrolles, J.-M., & Farines, V. (2016). Impact of

D

initial lipid content and oxygen supply on alcoholic fermentation in champagne-like musts.

PT E

Food Research International. https://doi.org/10.1016/j.foodres.2016.11.010 Otsuka, M., & Ohmori, S. (1992). Simple and sensitive determination of diacetyl and acetoin in biological samples and alcoholic drinks by gas chromatography with electron-capture

CE

detection. Journal of Chromatography B: Biomedical Sciences and Applications, 577(2),

AC

215‑ 220. https://doi.org/10.1016/0378-4347(92)80242-I Paria, S., Chatterjee, S., & Paine, T. K. (2014). Reactivity of an Iron–Oxygen Oxidant Generated upon Oxidative Decarboxylation of Biomimetic Iron(II) α-Hydroxy Acid Complexes. Inorganic Chemistry, 53(6), 2810‑ 2821. https://doi.org/10.1021/ic402443r Renouf, V., & Murat, M.-L. (2010). Gestion de la teneur en diacétyle des vins. Revue française d’oenologie, (240). Revel, G. de, Pripis-Nicolau, L., Barbe, J.-C., & Bertrand, A. (2000). The detection of ?-dicarbonyl compounds in wine by formation of quinoxaline derivatives. Journal of the Science of Food 19

ACCEPTED MANUSCRIPT and Agriculture, 80(1), 102‑ 108. https://doi.org/10.1002/(SICI)10970010(20000101)80:1<102::AID-JSFA493>3.0.CO;2-Y Romano, P., Brandolini, V., Ansaloni, C., & Menziani, E. (1998). The production of 2,3-butanediol as a differentiating character in wine yeasts. World Journal of Microbiology and Biotechnology, 14(5), 649‑ 653. https://doi.org/10.1023/A:1008804801778

PT

Romano, P., & Suzzi, G. (1996). Origin and production of acetoin during wine yeast fermentation. Applied and Environmental Microbiology, 62(2), 309.

RI

Ronkainen, P., Brummer, S., & Suomalainen, H. (1970). α-Hydroxy Ketones, Acetoin and Hydroxy

SC

Pentanone, in Wines. American Journal of Enology and Viticulture, 21(3), 136.

NU

Schmitt, H. D., Ciriacy, M., & Zimmermann, F. K. (1983). The synthesis of yeast pyruvate decarboxylase is regulated by large variations in the messenger RNA level. Molecular &

MA

General Genetics: MGG, 192(1‑ 2), 247‑ 252.

Sheppard. (2007). Vicinal Diketone Production and Amino Acid Uptake by Two Active Dry Lager

D

Yeasts During Beer Fermentation. Journal of the American Society of Brewing Chemists.

PT E

https://doi.org/10.1094/ASBCJ-2007-0515-01 Suomalainen, H., & Ronkainen, P. (1968). Mechanism of Diacetyl Formation in Yeast Fermentation. Nature, 220(5169), 792‑ 793. https://doi.org/10.1038/220792a0

AC

234‑ 235.

CE

van Neil, C. ., Kluyver, A. ., & Derx, H. . (1929). Über das Butteraroma. Biochemische Zeitschrift, (210),

White, F. H., & Wainwright, T. (1975). ANALYSIS OF DIACETYL AND RELATED COMPOUNDS IN FERMENTATIONS. Journal of the Institute of Brewing, 81(1), 37‑ 45. https://doi.org/10.1002/j.2050-0416.1975.tb03659.x

20

ACCEPTED MANUSCRIPT Figure legends: Figure 1: Metabolic pathways of vicinal diketones in yeast. Amino acids are boxed in green, diketones and their precursors in red, the products of glycolysis in black, and other intermediates or final products in blue. Dotted lines represent non-enzymatic reactions and dashed lines represent pathways without all the intermediates. Figure 2: Flowchart of sample preparation for total diketone and free diketone assay. Common steps are boxed in red and specific steps are boxed in blue.

RI

PT

Figure 3: Ratio of the peak areas between diacetyl and dodecane (a) and 2,3-pentanedione and dodecane (b) at five stages of fermentation (0%, blue diamonds; 25%, violet crosses; 50% red squares; 75%, orange circles; and 100%, green triangles).

NU

SC

Figure 4: Production dynamics of vicinal diketones and their precursors in a synthetic must: (a) in black, production rate of carbon dioxide (gCO2/L/h); in red, residual glucose concentration (g/L); in green, residual fructose concentration (g/L); (b) diacetyl (in blue), total diacetyl (in red), αacetolactate expressed in equivalent diacetyl (in green); and (c) pentanedione (in blue), total pentanedione (in red), and α-acetohydroxybutyrate in 2,3-pentanedione equivalents (in green).

MA

Figure 5: Production dynamics of vicinal diketones and their precursors in a natural must (pinot noir): (a) diacetyl (in blue), total diacetyl (in red), α-acetolactate expressed as diacetyl equivalents (in green), carbon dioxide production, in grey (gCO2/L/h); and (b) pentanedione (in blue), total pentanedione (in red) and α-acetohydroxybutyrate expressed as 2,3-pentanedione equivalents (in green), carbon dioxide production, in grey (gCO2/L/h).

PT E

D

Figure 6: Production dynamics of total diacetyl in natural must of three different pinot noirs: pinot noir n°1 (bleu squares), pinot noir n°2 (red squares), pinot noir n°3 (green squares).

Table legend:

CE

Table 1: Composition of natural musts of pinot noir

AC

Table 2: Data obtained from calibration range carried out at three levels of fermentation progress with dodecane and deuterated diacetyl as internal standards. Sample is diluted 5 times in a citrate phosphate buffer at pH 7 for each calibration range.

21

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

AC

CE

PT E

D

Fig. 1

22

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

AC

CE

PT E

D

Fig. 2

23

PT

ACCEPTED MANUSCRIPT

AC

CE

PT E

D

MA

NU

SC

RI

Fig. 3

24

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

AC

CE

PT E

D

MA

Fig. 4

25

PT

ACCEPTED MANUSCRIPT

AC

CE

PT E

D

MA

NU

SC

RI

Fig. 5

26

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

AC

CE

PT E

D

MA

Fig. 6

27

ACCEPTED MANUSCRIPT Table 1 : Composition of natural must of pinot noir Year

NTU

Ammonium mg/L

YAN mg/L

Total acidity g/L

pH

SO2 total mg/L

Volume hL

pinot noir (n°1)

2016

45

140

309

7.1

3.12

51

6

pinot noir (n°2) pinot noir (n°3)

2015 2014

51 75

58 143

211 276

6.6 7.4

3.08 3.06

32 39

23 120

AC

CE

PT E

D

MA

NU

SC

RI

PT

Grape Variety

28

ACCEPTED MANUSCRIPT

Table 2: Data obtained from calibration range carried out at three levels of fermentation progress with dodecane and deuterated diacetyl as internal standards. Sample is diluted Media g/L

Internal standard : Dodecane Diacetyl 2,3-Pentanedione

Gluco Fructo Ethan Intersecti Intersecti Slope R² Slope R² se se ol on on

Internal standard : Diacetyl-d6 Diacetyl 2,3-Pentanedione Slope

Intersecti Intersecti R² Slope R² on on

CE

PT E

D

MA

NU

SC

RI

0

AC

90

PT

0.50 0.99 0.33 0.99 0.32 0.99 0.34 0.99 -0.0137 -0.0038 -0.0164 -0.0027 62 98 5 98 9 89 27 93 0.48 0.99 0.36 0.99 0.32 0.99 0.33 0.99 45 45 42.3 -0.0052 -0.0125 0.0085 -0.0026 69 92 19 75 41 56 44 88 0.53 0.99 0.99 0.32 0.99 0.33 0.99 0 0 84.6 -0.0207 0.35 -0.014 -0.0005 -0.0079 16 98 59 91 38 72 08 0.50 0.34 0.32 0.33 Average 82 90 74 81 CV% 4.41 3.86 0.87 1.25 at 5th in a citrate phosphate buffer at pH 7 for each calibration range. 90

29

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

AC

CE

PT E

D

Graphical abstract

30

ACCEPTED MANUSCRIPT

Highlights

CE

PT E

D

MA

NU

SC

RI

PT

Chronology of synthesis of vicinal diketones in wine alcoholic fermentation Dynamic of production of α-acetohydroxy acids for wine alcoholic fermentation Adaptation and improvement of GC/MS method to oenological conditions Residual precursors constitute potential sources of vicinal diketones in wine

AC

   

31