“One-shot” analysis of wine parameters in non-Saccharomyces large-scale alcohol reduction processes with one- and two-dimensional nuclear magnetic resonance

Facing climate change in wine industry comprises the implementation of strategies, such as to reduce alcohol in wines, promoted by abnormal increment of sugar levels in wine grapes. The present work discusses the first industrial-scale use of specific yeast strains able to produce wine with reduced alcoholic concentration. Reduction of alcohol content and quantification of key metabolites associated to oenological practice and/or quality were simultaneously measured in a “one-shot” way with proton Nuclear Magnetic Resonance Spectroscopy. Novel relevant metabolites were revealed with the use of a two-dimensional 1H-13C HSQC multipresat correlation spectroscopy, whereas a detailed methodological NMR description is stressed, towards revealing novel resonances within the NMR signature. The use of multitask analytical methods to simultaneously describe alcohol reduction and NMR targeting, completes the portfolio of NMR solutions recently proposed to the World Organisation of Vine and Wine for as well quantify aging and varieties.


Introduction
Current oenological practices are focused in searching strategies for reducing alcohol content in wine, as climate changes has provoked an important increase of sugar amounts in must. With the simplest Saccharomyces yeast strains anaerobic conditions, fermentation of grapes with increased sugar content will produce wine with high alcohol by volume percentage (%ABV) and thus a product with penalized mouth-feel and/or taint and/or flavour [1,2]. In consequence, an important reduce of market consumption or severe tax government policies, could be expected. Amongst the broad spectrum of viticultural, physical or microbiological processes suggested for alcohol reduction, the concatenated use of non-Saccharomyces yeast strains to first aerobically sequester excess of sugar content by respiration, followed by the use of anaerobic Saccharomyces strains for final fermentation, has gained attention in the last years [3][4][5]. Methods comprising the use of several non-Saccharomyces yeast strains for assimilating must sugar to produce wine with reduced alcohol content and appropriate organoleptic properties has been extensively reported at lab-medium-and pilot-scales [6][7][8], but scarcely reported at industrial scale [9]. Finally, analysis of relevant wine parameters of reduced alcohol wines is carried out currently by a set of different enzymatic colorimetric trials and chromatographic schemes. Basic wine analysis comprises the quantification of primary metabolites such as ethanol, acetic acid, D-glucose, D-fructose, glycerol, lactic acid, malic acid and/or isovaleric acid -each with a particular enzymatic test kit-, that present some disadvantages such as time consuming; they present certain complexity in terms of sample preparation and chemical manipulations, they are costly and require some level of analytical expertise.
The present work proposes a "one-shot" evaluation of basic dealcoholized wine parameters with one-and two-dimensional Nuclear Magnetic Resonance (NMR) fingerprint and profiling [10][11][12]. Mono-varietal Mexican Cabernet Sauvignon wines with reduced alcohol content, were prepared in a large-scale regime (3.0 Ton grape, final volume of c.a. 2900 L) with non-Saccharomyces Candida zeplina strain, in co-inoculation with S. cerevisiae for final fermentation. Said reducing alcohol content procedure, was compared with at least two alcoholic fermentations at the same scale, carried out respectively with S. bayanux or S. cerevisiae (CM) strains. Ethanol reduction with simultaneous profiling and quantification of key metabolites such as acetic acid, malic acid, sorbic acid, fumaric acid and shikimic acid were carried out with a recent OIV procedure [10]. Finally, the use of 2D-NMR acquisition schemes are proposed in order to increase

Large scale fermentations
Large scale production of wine with reduced Alcohol content (%alc. v/v) was carried out in two different regions: Baja California, Mexico (Monte Xanic, hereinafter called as MX) and Parras, Coahuila, Mexico (Casa Madero, hereinafter called as CM), using the variety Cabernet Sauvignon, obtaining yields of c.a. 2950 L of wine, using as raw material 3.0 Ton Cabernet Sauvignon grapes. Prior to vinification, used grapes presented the following chemical properties, resumed in Table 1.
Large-scale alcohol reduction were done with the following non-conventional strains (Tables 2-4), whereas results will be compared against a large-scale control process (Saccharomyces bayanus, Strain 10N45).

Nuclear magnetic resonance (NMR) spectroscopy
Sample preparation for NMR studies comprised the addition of 100 µL of a mixture of D 2 O and chemicalshift reference sodium 3-(trimethylsilyl)-propionate-2, 2, 3, 3-d 4 (TSP), phosphonate buffer KH 2 PO 4 0.1% and 2% Oxygen need NE NaN 3 to 900 uL of wine sample, whereas pH was finally adjusted to a value of 3.9 for all samples. Samples were finally versed in standard 5 mm NMR tubes. All spectra were recorded on a Bruker 600 AVANCE III HD equipped with a 5 mm 1 H/D TXI probehead with z-gradient. The following set of NMR experiments were conducted at a temperature of 298 K, stabilizing the temperature with a Bruker VCU flow unit: a) Quantitative one-dimensional proton nuclear magnetic resonance spectra (q-1D-1 H-NMR) used to measure the reduction of alcoholic content in wines were carried out by recording a total of 64 transients, that were collected into 28,844 complex data points, with a spectral width of 20 ppm (12019 Hz), an optimized recovery delay of 5.6 seconds to obtain quantitative signal integration and acquisition times of 1.2 s, produced experimental times of 7 minutes per experiment. No apodization function was used prior to Fourier Transformation. b) 1D-1 H experiments with water-to-ethanol solvent presaturation were carried out as elsewhere reported [12]. c) Two-dimensional 1 H-13 C Heternouclear Single Quantum Coherence experiment [13] with a homemade water-to-ethanol multipresaturation scheme [12], prior to INEPT polarization transfer [14] were recorded by acquiring 7810 × 256 points with 32

Results and discussion
Q-1D-1 H-NMR spectra of full set of samples ( Fig. 1) were obtained in order to calculate %alc. (v/v) of each fermented product, by means of signal integration of methyl (δ = 1.026 ppm, triplet)-to-methylene (δ = 3.497 ppm, quartet) signals, with respect water signal (δ = 4.7 ppm, singlet). Direct percentage of integrated methylene signal (δ = 3.497 ppm, quartet), with respect integrated H 2 O signal (δ = 4.7 ppm, singlet), immediately provides the alcohol percentage of fermented samples, as methylene and water signals present both the same number of observed spins (I = 2). Alcohol reduction of each large scale trial, computed by q-1D-1 H-NMR was of around 1% in all cases (see Fig. 3  by the PUlse Length based CONcentration (PULCON) method [15]. Detection and quantification of acetate, malate, sorbate, fumaric and shikimic moieties from the NMR signature is related to the easiness for signal selection of isolated-intense resonances from above mentioned metabolites, even exposed within the q-1D-1 H-NMR spectra, without multipresat scheme (Fig. 1), but at low signal-to-noise ratio. However, novel schemes involving solvent elimination and addition of a second dimension for dispersing encumbered resonances from the 1D-1 H fingerprint will shed light in novel resonances or exposed regions, potentially ready to be quantified with PULCON/NMR procedures. For that it is presented the use of a two-dimensional 1 H-13 C Heternouclear Single Quantum Coherence (HSQC) experiments, applying a home-made water-to-ethanol multipresaturation scheme [12] for assigning novel resonances involving relevant metabolites associated to oenological practices or quality control parameters.  Set of novel assigned resonances with the use of twodimensional 1 H-13 C HSQC are: tyrosine, phenylalanine, fructose, glycerol, glutamine, lactic acid and quercetin. Assignments were done by confirming a 1 H-resonance with its correlation to a specific 13 C chemical shift. It is worth noting to highlight as example that quercetin proton resonance (6.68 to 6.77 ppm) present a shift as a function of large-scale fermentation scheme. Despite the last, the easiness to identify quercetin relies on the unambiguous carbon correlations (δ 13 C = 116 and 137 ppm) with shifted protons, proving though the advantages of using additional NMR dimensions.

Alcohol reduction with respect controls
Quantification with PULCON/NMR method consists in referring the signal integral of an unambiguously assigned resonance (or set of them), with respect signal intensity and line width of a known external reference, prepared at chemical conditions close to the sample of unknown concentration. Set method has been proposed  for proteins [15] and recently to wine targeting [10], whereas for the later study, 20 mg/L of citric acid was used as external reference [10]. For the present study a set of external references were used for cross-check validation: 650 mg/L of acetic acid; 1436 mg/L of malic acid; 152 mg/L of sorbic acid; 49 mg/L of fumaric acid and 51 mg/L of shikimic acid. All external reference solutions were prepared at similar pH and buffer conditions with respect wine samples (see Materials and Methods). PULCON/NMR targeted analysis is stressed in Fig. 5 and Table 5.

Conclusions
A multitask NMR strategy is herein proposed to analyse alcohol reduction and oenological practice in wines produced with the first large-scale Non-Saccharomyces or non-conventional Saccharomyces fermentation processes. Standard metabolites were detected and quantified with the PULCON/NMR procedure and quality recommendations can be done as a function of selected fermentation scheme for reducing %alc. (v/v). High-resolution two-dimensional 1 H-13 C HSQC with solvent multipresat have revealed novel resonances ready to be targeted and quantified with PULCON/NMR method, towards a not known accuracy in wines' quality control.